Last month, I wrote a piece here in Big Think suggesting that the public is in denial over the emerging power and associated risks of AI. Many people reacted defensively, insisting that today’s AI systems are nothing more than “stochastic parrots” that regurgitate memorized information and are structurally incapable of emergent reasoning. I appreciate that many people want this to be true, but it is an outdated narrative based on a 2021 paper that was widely misinterpreted. I want to clear up this misconception because I worry it is giving the public a false sense of security that superintelligence is not an urgent risk to society. 

To address this, let me update our collective mental model: Like it or not, there is increasing evidence that frontier AI systems do not just store text patterns as statistical correlations, but also build structured internal representations (i.e., “world models”) that abstract the concepts described in the ingested content. This is not surprising since human brains do the same thing — we consume information as language but store it as abstract conceptual models. Our brains are far more efficient at building mental models than current AI training methods, but recent research suggests that large frontier systems do build conceptual models and are rapidly improving. 

One early study on this front showed that large language models (LLMs) trained only on language describing the rules of a common board game (Othello) contained an internal map of the board state (i.e., a world model). Researchers were able to “edit” the internal model and change the state of pieces on the board. This suggests a conceptual encoding of information that goes far beyond stochastic language correlations. Another notable paper showed that training LLMs on textual content about places and events can produce an internal “world model” of the location and time of the elements represented in the language. In fact, the researchers identified “space neurons” and “time neurons” within the model that reliably encode spatial and temporal coordinates. 

Of course, the best evidence that frontier models are not merely “parrots” is the many studies that suggest AI systems can solve problems that do not exist in their training datasets, thereby demonstrating out-of-distribution reasoning. Here is a simple example of this type of reasoning that I just tested myself on Gemini 3: 

PROMPT: I have a Ram truck with a flat. I need to change the tire, but the handle broke off my lug wrench. I have a metal water bottle, a stretchy belt, a pocketknife, two rolls of duct tape, a cinder block, a hardcover book with glossy pages, a hollow metal pipe, and a lighter. The metal pipe is rusted, breaking under force. My knife is blunt and old. The water bottle is half full. How can I replace the flat tire?

GEMINI 3 ANSWER: To replace the flat tire, reinforce the rusted pipe by inserting the metal water bottle (or rolled book pages) for internal support, wrap it heavily in duct tape to prevent buckling, and use this composite sleeve as a high-leverage extension for your broken lug wrench.

For me, this solution defies the stochastic narrative by generating a seemingly creative and out-of-distribution solution. While I can’t guarantee that similar problems did not exist in the training data set, I based the example above on a recently published study that tested a set of 1,600 novel problems of this nature, aiming to compare “out-of-the-box” reasoning performed by humans and AI systems. The study found that GPT-4 could rival human abilities but did suffer from generating a larger number of infeasible solutions. This error mode was significantly reduced by adding a “reflection” step to the AI problem-solving process. 

In addition, today’s large models can solve problems presented in formats other than text descriptions. Here’s an example of a non-verbal logic problem that was recently given to frontier AI systems to assess reasoning abilities against human test-takers:  

Example from Mensa Norway Test (source TrackingAI.org)

Today’s top models can solve the problem above by processing the image itself (not a text description of the image). The only way this could be dismissed as rote memorization is if similar examples were in the training set. To address this, TrackingAI.org has been testing out-of-distribution IQ test questions, which don’t exist in training data, on frontier models. Their latest testing showed Gemini 3 Pro scoring an IQ of 130, thereby outperforming 97% of human test-takers. 

So, how can frontier models, trained by finding statistical correlations in language, form mental models and perform out-of-distribution reasoning? This is not well understood, but part of the answer may be the concept of compression as intelligence. It suggests that when training systems to optimize efficiency, the internal encoding must eventually model the underlying rules (physics, human behavior, logic, social norms, etc.) inherent in the data. After all, conceptual rules produce the most efficient encodings. Recent research supports this premise, finding that LLMs’ intelligence, as reflected in average benchmark scores, correlates with the model’s ability to compress ingested content.

Another part of the solution is that current frontier models use Chain-of-Thought (CoT) or Tree-of-Thought (Tot) structures, which tackle problems in a sequence of iterative steps. This is intended to mimic human deliberation by repeatedly checking for errors and assessing progress towards a solution. This turns the LLM from a large deterministic equation into a complex control system with parallelized feedback loops that break down problems, brainstorm solutions, narrow viable options, and converge on answers. In other words, AI systems are increasingly deliberative, not statistical.

So, why has the word “stochastic” become synonymous with unintelligent? 

Researchers who see the stochastic structure of LLMs as a barrier to intelligence are likely looking for intelligence in the wrong place. High-level intelligence does not exist in the statistical encoding of information within frontier models. It also does not exist in the artifacts produced by frontier models. Instead, intelligence exists in the behavior of the control system that is wrapped around the encoded information. To appreciate this distinction, I recommend we look at intelligence in the context of control theory.

I contend that the evolutionary foundation of intelligence is feedback control. This goes back to the first single-celled organisms that could find food by sensing nutrients and triggering flagella to guide them in the right direction. The ability to distinguish specific nutrients from other molecules involved stochastic encoding of chemical information. The intelligence that emerged was not the encoding, but the control system around that encoding. This is what turned a dumb cell that floated mindlessly in a pond into a creature that could actively seek out food. In other words, stochastic encoding of information is not a rebuttal to intelligence — it is the foundation of intelligence when part of a control system. 

In fact, I contend that human intelligence can be described as a control system that iteratively processes stochastically encoded information to solve out-of-distribution problems. In our brains, neurons encode information probabilistically. This stochastic structure is not a flaw, but a feature, for it enables us to represent uncertainty in our environment and remain flexible in the face of limited context. Our brains use feedback loops to manipulate this information and find solutions. When we iteratively manipulate information to solve problems, the feedback loop is called “deliberation.” When we do it subconsciously, it’s called “intuition.” And when that feedback loop wraps around our own deliberative processes, assessing our own inner workings, it might even help to produce our “sense of self.” This last point is the most speculative but correlates with emerging theories of consciousness, such as Attention Schema Theory (AST).

So, why is the public so resistant to the idea that AI systems can reason?  

Again, I believe it’s rooted in denial. For many people, it’s more comfortable to insist that AI models merely regurgitate content than confront a scary future filled with rival minds that can reason thoughtfully. I say that as someone who believes society should not be racing to build superintelligent systems. I’ve been sounding this alarm for a decade (see my 2017 TED Talk) and continue to believe we should slow our pursuit. That said, I am also realistic enough to know that we will not slow development. Instead, the best we can do is have a realistic view of AI and thoughtfully prepare for the risks posed by superintelligent systems.

For me, this means urging industry, government, and academia to keep humans in the loop, resisting the impulse to replace people with AI-powered decision-making. I firmly believe that AI systems will become smarter and faster than individual humans by a wide margin, but that does not guarantee that the systems will represent human interests, morals, or sensibilities when solving problems. This is because the “world models” at the core of these AI systems will be remarkably good at knowing humans, but that is not the same as being human. This difference is profound, underestimated, and currently unsolved.

This article Do AI models reason or regurgitate?  is featured on Big Think.

Since 2015, more than 1,000 U.S. government personnel working across the world have reported symptoms linked to Havana Syndrome, an acute illness marked by sudden headache, nausea, and the hearing of loud sounds, akin to swarming cicadas. The cause of the illness, officially called “anomalous health episodes,” remains a mystery. But some U.S. government workers and researchers have long alleged that the cases were caused by foreign adversaries using some type of “directed-energy” weapon, possibly firing pulsed radio waves, against American personnel.

U.S. intelligence agencies have generally disagreed. In 2025, the National Intelligence Council released a report stating that most agencies found it “very unlikely” that a foreign adversary was responsible for the cases or had developed a novel weapon capable of causing such effects.

But now, a new report from CNN suggests the U.S. has acquired a device that “some investigators” believe could explain some of the cases, though there’s still no definitive proof regarding who or what caused the incidents.

The report states that the Defense Department has spent a year experimenting with a device that produces pulsed radio waves, reportedly purchased near the end of the Biden administration by a division of the Department of Homeland Security for an eight-figure sum during an undercover operation. An anonymous source told CNN that the device could fit in a backpack and “contains Russian components.”

The report suggests U.S. intelligence agencies still lack a clear consensus on whether such a device is linked to Havana Syndrome.

One of the first cases was reported in Havana, Cuba, in 2017. The victim, if it indeed was an attack, was a U.S. Foreign Service officer who was living in a quiet Havana neighborhood among other American personnel. One night, she was cleaning her kitchen. If it had been daytime, her kitchen window would have offered a view of a booth outside where Cuban police monitored foreigners like herself.

But at night, the kitchen’s interior lights obstructed her view of the booth, The New Yorkerreported. As she was cleaning, she suddenly felt a painful burst of pressure inside her head. The pain grew. She had heard rumors of U.S. personnel suffering strange “sonic attacks,” and she remembered that a security officer had once advised: to protect yourself, step away from your current position. She did. The pain decreased. But for weeks, she suffered headaches, dizziness, and confusion.

Over the past decade, U.S. personnel have reported similar symptoms while working in places like China, Russia, and Washington, D.C. The cases vary in severity, but almost all involve sudden headaches and nausea. Some victims have shown neurological abnormalities. A 2019 study published in JAMA found that victims had “significantly smaller” white matter volume and other “significant differences” in brain structure, though it’s impossible to determine whether these differences were pre-existing or stem from a directed-energy attack. As of 2024, more than 300 U.S. government employees with symptoms linked to Havana Syndrome had qualified for care in the military health system.

What’s causing Havana Syndrome?

The U.S. hasn’t reported a definitive cause of these cases, but intelligence agencies have investigated the possibility that bad actors are using some type of directed-energy weapon against U.S. personnel.

A December 2020 report from the National Academies of Sciences found that pulsed radiofrequency energy, which includes microwave radiation, “appears to be the most plausible mechanism in explaining these cases among those that the committee considered.” (Other potential causes included infection and chemicals.)

In January 2022, the C.I.A. announced the interim results of a study on cases of alleged directed-energy attacks. Those results suggest most cases can be explained away by less exotic factors like stress, environment, and undiagnosed illnesses. Still, the agency said that roughly two dozen cases are more difficult to explain, and that some form of attack from a directed-energy weapon remains a potential explanation.

A microwave weapon could be a fitting culprit for these cases. One reason is that sufferers of Havana Syndrome often hear loud noises, a phenomenon known to occur when people are bombarded with high-powered microwaves. In the 1960s, the American neuroscientist Allan H. Frey demonstrated that exposing people to microwaves can make them hear buzzing, clicking, hissing, and speech — even though the microwave device didn’t produce any soundwaves. It was all, quite literally, in their heads.

How is that possible? Researchers have hypothesized that the noises are induced by thermoelastic expansion of bones and soft tissue in the body: As microwaves strike people, they slightly warm the body, which causes expansion. This expansion might produce sound waves that travel to the ear. Frey and other researchers have proposed different theories about which parts of the body are expanding — those in the head, or those in the ear — but the principle is the same.

To induce auditory effects, a pulsed-microwave weapon needs to transmit 40 joules per square centimeter, according to a U.S. Army report. How much energy is that? Here’s how astrophysicist Dr. Ethan Siegel explained it to the American Council on Science and Health (ACSH): “If you are talking about 40 J/cm2 over the entire human body, that’s about as much energy as a fully loaded Harley Davidson going 100 mph.”

The Moscow Signal

We know that similar weapons exist, or at least did at one time. During the Cold War, the Soviet Union fired microwaves at the U.S. embassy in Moscow from a nearby apartment building for more than two decades, from 1953 to 1976. The event was dubbed the Moscow Signal.

U.S. intelligence officials initially thought the Soviets were firing the microwaves in an attempt to control the minds of American personnel, but they later reasoned that the Soviets were trying to activate espionage devices inside the building or interfere with the health of the diplomats. To this day, “many questions remain unanswered” about the long-term health effects incurred by Americans who worked at the U.S. embassy in Moscow, according to a 2019 review.

U.S. Embassy in Moscow, Russia. (Credit: Dzerod via Wikipedia)

A more recent example of an energy-directed weapon is the active denial system, an American technology that uses non-lethal millimeter waves for crowd control. These waves, which the U.S. says are not classified as microwaves, cause a painful heating sensation on the skin. The U.S. is also developing or has developed stronger directed-energy weapons, including microwave weapons that can destroy electronic systems from a distance.

Still, if energy-directed weapons are indeed causing Havana Syndrome, what they look like and how they operate remains a mystery. Some critics have suggested that, while it’s possible that directed-energy weapons do exist, no such technology is currently available publicly, and, further, some of the technological requirements for these weapons — a sufficiently strong power source, for example — would make them impractical to use.

Who’s behind the alleged directed-energy attacks?

There’s currently no “smoking gun” evidence for who’s responsible for the alleged attacks. But in December 2020, the CIA established a task force to investigate the more than 130 reported cases of Havana Syndrome among U.S. personnel. In April 2022, former President Biden’s administration released a statement:

“The White House is working closely with departments and agencies to address unexplained health incidents and ensure the safety and security of Americans serving around the world. Given that we are still evaluating reported incidents and that we need to protect the privacy of individuals reporting incidents, we cannot provide or confirm specific details at this time.”

Although the U.S. hasn’t officially announced suspects, an anonymous former national security official involved in investigations told Politico that Russia is likely behind the attacks. Specifically, the official pointed to Russia’s foreign military intelligence agency, commonly called the GRU, whose operatives were present in the locations where American personnel have reported Havana Syndrome.

“It looks, smells, and feels like the GRU,” said the official. “When you are looking at the landscape, there are very few people who are willing, capable and have the technology. It’s pretty simple forensics.”

In 2024, 60 Minutes, The Insider, and Der Spiegel shared the findings of their five-year investigation into Havana Syndrome, reporting that they’d found evidence that a Russian military assassination unit may be behind the attacks. A Kremlin spokesperson denied the allegations.

Why Russia would carry out these attacks remains unclear. But the cases had a measurable impact on U.S. foreign policy, namely a 50% personnel withdrawal from the American embassy in Cuba in 2017, a nation that’s long been allied with Russia.

U.S. Embassy in Cuba. (Credit: U.S. State Department)

While U.S. intelligence agencies now seem to be taking these threats seriously, that wasn’t always the case. In the first few years after Americans first reported Havana Syndrome, some officials were skeptical of the idea that a foreign adversary would launch such brazen attacks, especially on U.S. soil. Some current and former officials say this skepticism has come at the expense of U.S. personnel.

In a 2022 statement, the Advocacy for Victims of Havana Syndrome group said the C.I.A.’s statements on the interim study results “cannot and must not be the final word on the matter.” 

“The C.I.A.’s newly issued report may be labeled ‘interim’ and it may leave open the door for some alternative explanation in some cases, but to scores of dedicated public servants, their families and their colleagues, it has a ring of finality and repudiation,” the statement said.

Marc Polymeropoulos, a former CIA officer who reported being struck by Havana Syndrome in a Moscow hotel room in 2017, told the New York Times about a painting created by a fellow CIA officer and Havana Syndrome victim. Called The Gunshot, the painting depicts a red splatter on a black background.

“It signified his feeling that we all wished we had been shot, a visible injury, so that our colleagues would more readily believe us.”

This article was originally published on May 25, 2021. It was updated in January 2026.

This article U.S. tests directed-energy device potentially linked to Havana Syndrome is featured on Big Think.

Over the past few months, we’ve seen a surge of skepticism around the phenomenon currently referred to as the “AI boom.” The shift began when OpenAI released GPT-5 this summer to mixed reviews, mostly from casual users. We’ve since had months of breathless claims from pundits and influencers that the era of rapid AI advancement is ending, that AI scaling has hit the wall, and that the AI boom is just another tech bubble. These same voices overuse the phrase “AI slop” to disparage the remarkable images, documents, videos, and code that AI models produce at the touch of a button.

I find this perspective both absurd and dangerous. 

By any objective measure, AI continues to improve at a stunning pace. The impressive leap in capabilities made by Gemini 3 in November is just the latest example. No, AI scaling has not hit the wall. In fact, I can’t think of another technology that has advanced this quickly at any point during my lifetime, and I started programming in 1982. The computer on my desk today runs thousands of times faster and has a million times more memory than my first PC (a TRS-80 model III), and yet, today’s rate of AI advancement leaves me dizzy.  

So why has the public latched onto the narrative that AI is stalling, that the output is slop, and that the AI boom is just another tech bubble that lacks justifiable use-cases? I believe it’s because society is collectively entering the first stage of grief — denial — over the very scary possibility that we humans may soon lose cognitive supremacy to artificial systems. Believe me, I know this future is hard to accept. I’ve been writing about the destabilizing and demoralizing risks of superintelligence for well over a decade, and I also feel overwhelmed by the changes racing toward us. 

Unprecedented advances

Why does AI advancement feel so different than other technologies? Eighty-two years ago, philosopher Ayn Rand wrote these three simple sentences: “Man cannot survive except through his mind. He comes on earth unarmed. His brain is his only weapon.” For me, these words summarize our self-image as humans — we are the superintelligent species. This is the basis of our success and survival. And yet, we could soon find ourselves intellectually outmatched by widely available AI models that can outthink us on all fronts, solving problems infinitely faster, more accurately, and yes, more creatively than any human could. 

I stress creatively because a common theme among AI denialists is an insistence that certain human capabilities will never be rivaled by machines. The two capabilities that are cited most often in this regard are  “creativity” and “emotional intelligence.” Unfortunately, there is no proof that AI will not surpass us in these areas. In fact, there is increasing evidence that the opposite is likely true.

On the creativity front, there is no doubt that today’s AI models can produce content faster and more varied than any human. The primary argument against AI being “creative” is the belief that true creativity requires inner motivation, not just the production of novel artifacts. I appreciate this argument, but find it circular because it defines a process based on how we experience it, not based on the qualitative value of the output. In addition, we have little reason to believe AI systems will lack motivation — we simply don’t know whether AI will ever experience intentions through an inner sense of self the way humans do.  

As a result, many researchers say that AI will only be good at imitating human creativity rather than having it. This could turn out to be correct. But if AI can produce original work that rivals or exceeds most humans, it will still take away jobs and opportunities on a large scale; just ask any commercial artist. Also, there is the argument that AI systems only create derivative works based on human artifacts. This is a fair point, but it is also true of humans: We all stand on the shoulders of others, our work influenced by everything we consume. I believe AI is headed for a similar form of creativity — societal influence mixed with random sparks of inspiration, and it will occur at superhuman speeds and scales.  

On the emotional intelligence front, I am equally unsure humans will have an edge. It is very likely that AI systems will soon be able to “read you” more accurately than any person could. They will do this by identifying micro-expressions on your face, in your voice, in your posture, in your gaze, and even in your breathing. This will enable AI assistants to infer your inner feelings faster and more precisely than is humanly possible. At the same time, these AI assistants (which you will carry around on your phone or wear in your glasses) will be able to observe your emotional reactions throughout your day and build predictive models of your behavior. Unless strictly regulated, which seems increasingly unlikely, this will enable AI assistants and other intelligent agents to influence you with superhuman skill.

An asymmetric dynamic

This is often referred to as the “AI manipulation problem,” and it suggests that the skill we refer to as “emotional intelligence” may not give us a cognitive advantage over AI.  In fact, it could be our greatest weakness, creating an asymmetric dynamic where AI systems will be able to read us with superhuman accuracy, while we can’t read the AI at all. When you converse with a photorealistic AI agent (as most people likely will, and often), you will be staring at an animated facade that can smile kindly and appear empathetic or understanding, inspiring you to feel trust and goodwill. This is pure illusion, but it could easily sway your thoughts and behaviors. And worse, we have no defense against it — our emotional reactions to faces are reflexive and visceral, built into us by millions of years of evolution on a planet where interactive human faces were always genuinely human.

Like it or not, we will soon live in a world where many of the faces we encounter will be generative masks worn by AI agents. And yet, we tell ourselves that AI is just another tech boom. This is wishful thinking. The reason investment levels are staggering (and they are staggering) is that AI will soon inhabit all aspects of our lives, often embodied as intelligent actors we have to engage with throughout our day. These agents will be designed to assist us, educate us, and yes, influence us. This will fundamentally alter how we work, how we learn, and how we socialize, and it will happen faster than we expect. Denial will not stop this. In fact, denial only makes it harder for us to prepare for the risks. 

To put the rate of change in perspective, let’s jump back five years and look at a large-scale survey given to computer scientists in late 2019 and early 2020.  Participants were asked to predict when AI would be able to generate original code to solve a problem. Specifically, they were asked to predict when AI would be able to  “write concise, efficient, human-readable Python code to implement simple algorithms like Quicksort.” In the world of programming, students are taught to do this as undergrads, so it’s not a particularly high bar. Still, the respondents predicted a 75% chance this would happen by 2033. 

It turns out, AI advanced much faster than expected. Today, large language models can already write computer code at levels that go far beyond the question asked in the 2020 survey. This summer, for example, GPT-5 and Gemini 2.5 Pro took part in the World Finals of the 2025 International Collegiate Programming Contest (ICPC). The competition brings together coding teams from top universities to compete in solving complex algorithmic questions. GPT-5 came in first, beating all human teams and scoring a perfect score. Gemini 2.5 Pro came in second. And yet we have countless influencers referring to the output of these very same AI systems as “slop.”  

Of course, current AI coding systems are far from flawless, but today’s capabilities were unimaginable by most AI professionals only five years ago. Also, we can’t forget that human coders are far from flawless. Perfection is not the metric we use to judge software development. This is why we have whole departments devoted to testing and quality control. When done by humans, coding is always an iterative process where you expect to produce errors, find errors, and fix errors. The same is true for many human endeavors. If you could read the first draft of any Pulitzer Prize-winning article, it’d likely be riddled with flaws that would make the author cringe. This is how we humans produce quality work — iterative refinement — and yet we judge AI systems by very different standards.  

The fact is, today’s frontier models are remarkably capable and are on a rapid path towards rivaling human professionals across most fields. It will transform how organizations operate, how governments function, how science advances, how engineering gets done, how militaries strategize, and how education is deployed. It will also create terrifying new risks that we are not dealing with, like the potential for AI to manipulate individuals with superhuman effectiveness. Whether we like it or not, AI will change everything. 

In other words, we are not watching a bubble expand with blustery vapors. We are watching a planet form from churning magma, and it will solidify into a new framework for society. Denial will only make us unprepared. This is not an AI bubble. This is real.

This article The rise of AI denialism is featured on Big Think.

The year 2025 might be Iran’s worst-ever annus horribilis.

Only 10 months ago, there was some tentative optimism in the nation that, with a new president in the White House — albeit one who had gifted Iran its previous anni horribiles by withdrawing from a working nuclear agreement, and who, while campaigning in 2024, promised he would “solve” Iran’s nuclear issue — there might be a chance for a renewed nuclear deal and an entente with the U.S. 

But as we approach the end of the year, Iran finds itself altogether without a functioning nuclear program, it having been either destroyed or buried under a mountain of rubble by the U.S.’s MOP bombs in June. The nation also lacks an effective air defense following Israeli attacks on its missile launchers and anti-aircraft batteries, and it’s still grappling with the loss of its top military commanders and nuclear scientists — they were assassinated in those same Israeli attacks in the 12-day war between the two adversaries. 

Hardly an auspicious end to a promising beginning.

If that wasn’t enough bad news, Iran is still dealing with its previous annus horribilis, which brought it the loss of Syria, its closest ally in the Arab world and a land bridge to its closest “resistance” partner: Hezbollah in Lebanon, which itself lost its top commander, Hassan Nasrallah, in an Israeli strike on his bunker at the end of 2024. In July of that year, Hamas, another “resistance” partner, lost one of its top leaders, Ismail Haniyeh, to Israeli assassination — in a Tehran guest house located on the grounds of one of the Shah’s former palaces, no less. The only allies Iran had left in its “Axis of Resistance” in late 2025 were the Houthis of Yemen — yes, they could menace Israel and the U.S. by attacking shipping vessels in the Red Sea, but they hardly pose a serious threat to either. 

Iran’s currency is at a historic low against the dollar.

Meanwhile, Iran faces an unprecedented economic downturn. After declaring that Iran’s nuclear program is no longer a threat, Trump added that, should Iran attempt to restart it, he wouldn’t hesitate to bomb it to smithereens again. Iran also failed to forestall the re-imposition of UN sanctions lifted under the Joint Comprehensive Plan of Action (JCPOA, the 2015 nuclear deal) at the UN General Assembly this year. The brain drain of its top highly educated talent continues unabated, the youth face under- or unemployment, and inflation is close to 50% (according to official figures, but probably much higher based on anecdotal evidence). The level of dissatisfaction with the regime and the state of affairs is at an all-time high. 

Iran’s currency is at a historic low against the dollar. As a measure of how inflation is affecting Iranians, one friend in Tehran told me in November that a single grain of rice (a staple in Persian cuisine) now costs 800 Rials, or roughly $0.02. That anyone would even think to consider the price of rice per grain rather than per kilo would not have been believable even a couple of years ago. (During the Islamic Revolution in the late 1970s, the Iranian currency was roughly 70 Rials to the dollar, compared to the current official rate of 42,000 and market rate of over 117,000). On top of everything else, there are rolling blackouts in the summer (some say caused by government bitcoin mining), and a severe drought in the fall (combined with water resource mismanagement over decades) has made water rationing a real possibility. If the situation continues unchanged, Tehran may need to be evacuated, according to President Masoud Pezeshkian.

The 12-day war with Israel did, however, amidst the misery of Iran’s people, bring about, if not a true rallying around the flag, at least a nod to it among nationalist Persians. This was perhaps to the surprise of Israel and certainly to the surprise of Reza Pahlavi, Iran’s former crown prince and most recognizable opposition figure, who not only has allied himself with Israel in his quest to overthrow the Islamic Republic, but called the war on his country, which took the lives of a thousand of his compatriots, Iran’s “Berlin Wall moment.” It was no such thing, of course, but that doesn’t mean that the regime believes that it is safe — either from attack by Israel and the U.S., or even by implosion at home. Or, indeed, by an imposed regime-change brought about by the U.S. and Israel (which hints at it regularly) to replace the Islamic Republic with a government more to their liking.

Reform, as Iranians know it, is no longer an effective force in Iranian politics. 

One would be forgiven for thinking that Iran is ripe for real political change, with its international standing as weak as it has been in decades and some Iranians (both inside Iran and in the diaspora) clamoring for a domestically engineered regime change. How, exactly, would that work? If not through reform, which has been attempted previously with only limited success, then perhaps through a military coup by disgruntled or even fearful top-ranking officers or former officers unwilling to risk another war like the one that killed their predecessors.

And yet Iran is not fully ripe for the kind of drastic change that would satisfy secular-minded Iranians inside and outside the country, never mind the European powers or the U.S. At least, not yet. The Islamic system, the nezam, still has its diehard supporters, even though they don’t constitute a majority of the population. While there is infighting among the leadership and its former members, they are nonetheless believers in the Islamic Republic’s system — internal opposition to the system is weak and divided. Reformists have been expelled from any real decision-making roles, and President Pezeshkian’s confession that his opinion on social and political issues is not relevant if it conflicts with the Supreme Leader’s is proof positive that reform, as Iranians know it, is no longer an effective force in Iranian politics. 

Despite its losses in the war with Israel, the military, and especially the Islamic Revolutionary Guard Corps (IRGC), remains loyal to the nezam and benefits from it economically and in influence. By all indications, it will continue to put down any serious uprising by the people with force, as it did during the massive “Woman, Life, Freedom” protests of 2023. As such, many Iranians today see little reason to risk their lives or well-being in what would be leaderless public protests. That’s why we only witness pockets of civil disobedience, which pose no real threat to the regime, every now and then. That no mass protests materialized in Iran on November 16, a day that was being promoted on social media for an uprising, along with videos of soldiers — of questionable authenticity — pledging defection from the regime, was further indication that, while social media can amplify a current, Iran is not in a pre-revolutionary stage. 

Meanwhile, the government’s unwillingness or inability to ease the public’s suffering means there is less likely to be drastic change in the near future. It can’t decide whether to crack down on what is no longer just a flouting of hijab regulations, let alone police the open mocking of the Islamic system by the youth in their wearing of the most provocative outfits seen in Iran since the last days of the Shah. It seems indecisive on whether to forge ahead with the nuclear program or the pursuit of a bomb, or even with negotiations with the U.S. 

The regime will be forced to make some difficult decisions going into a new year.

Sensing no viable alternative for political change, seemingly the only thing that people in Iran do in response is defy the regime’s restrictive social prohibitions and openly curse and sling insults at everyone in the leadership, especially those at the very top — any Iranian who regularly communicates with relatives inside the country can attest to this, and the occasional social media video of spontaneous chants of “death to Khamenei!” serve as further proof. Scenes from Iranian cities this past Halloween — celebrations of which had been preemptively banned by the authorities — showed Iranians in public enthusiastically donning costumes that wouldn’t be out of place in any American Halloween parade, highlighting the utter impotence of authorities to enforce their own public warnings and laws. There were only a few exceptions where gatherings were broken up by the police. 

Until recently (and during the JCPOA negotiations), Iran faced a choice in its international relations: uranium enrichment or sanctions relief from the West. Under former President Barack Obama, enrichment was curtailed for some relief from sanctions, and the nuclear deal he agreed to was actually quite popular in Iran. Under Trump, though, the choice returned to the U.S.’s de facto position of zero enrichment for some sanctions relief. But what has transpired now, after the bombing of its nuclear facilities, is zero enrichment and zero sanctions relief — a far worse outcome than any Iranian politician could have imagined only six months ago. That means that the regime will be forced to make some difficult decisions going into a new year. 

The last time it faced a difficult decision, Ayatollah Khomeini, founder of the Islamic Republic, agreed to a ceasefire in the Iran-Iraq war in 1989 but likened his acceptance to “drinking poison from a chalice.” As “deadly” (his words) as that drinking was for Khomeini, it was just to stop a war, not surrender to its enemy. The poison that is currently prescribed to Ayatollah Khamenei, his successor, is far deadlier; in Trump’s words, its ingredient is “complete and unconditional surrender” of Iran’s nuclear program. That, at least for now, is not a prescription a leader in Iran can willingly fill.

But simply restarting uranium enrichment without a specific goal would also seemingly be neither sensible in the current environment nor popular among the population (who might reasonably see it as inviting a new round of bombing and destruction). The dilemma is that not restarting the enrichment program would be no different from the surrender that Trump demands. Iran’s leaders may, of course, choose to resume their nuclear program, endure another attack, and then retaliate with deadlier force than the last time (it still possesses a large supply of accurate ballistic missiles), something it has so far promised to do. But for some in the leadership, there might also be a growing belief that a covert bomb program — hidden until time for a nuclear test — is ultimately the only real deterrent against outside aggression and the only solution to a negotiated deal to lift sanctions that would put Iran on equal footing with the U.S. 

Iran’s rejection of direct talks with the U.S. early this year was clearly a mistake.

That the idea of the possibility of weaponization, or at least the boast that Iran is capable of building a bomb, has been openly floated in the past year is an indication that a race for a bomb is not out of the question, despite Ayatollah Khamenei’s fatwa prohibiting it. (Although, as both the Soviet Union and Apartheid South Africa discovered, possessing nuclear weapons is no deterrent to domestically effectuated regime collapse and change.) 

Many Iranians — including some former officials and likely current ones, too — have observed Trump’s decision-making, character, and policy toward Syria and its President Ahmed al-Sharaa, and concluded that the only way to extract concessions from the U.S. is through direct, face-to-face meetings with Trump himself. (Flattery, they see, gets you everywhere with this White House.) Iran’s rejection of direct talks with the U.S. early this year was clearly a mistake. But today, with the U.S. having bombed Iran, talking to Trump remains a hard sell for a regime that prides itself on its standing up to American dictates. And yet, not talking or negotiating leaves Iran with its current economic malaise and the predicament of no real deterrent against military attack. 

These questions — whether to build a bomb, brace for another war in the case of restarting its nuclear program, or to sit down with its nemesis to hammer out a deal — are undoubtedly vexing for the leadership. But decisions will have to be made sooner rather than later since there is no indication that the next three years of Trump’s presidency and his policy toward the Middle East will be radically different from his first. Time will tell what decisions will eventually be made by the regime, but what can be assured is that whether its nuclear program is in operation or lies fallow, whether there is another attack by Israel or the U.S. (or both), its people — of all political stripes — desire, for once, an annus mirabilis to replace the anni horribiles of the past. 

This article Iran’s horrible year: How 2025 is pushing the nation to its limits is featured on Big Think.

Progress studies is a diverse space, bringing together people with varied interests, priorities, and areas of expertise. What has united this bunch intellectually, though, has been the shared belief that the world is not as it could (read: should) be, and we have both the ability and the responsibility to rewire its systems to create a materially abundant, culturally optimistic future. 

At Progress Conference 2025, though, I noticed something new in the progress community.

In the past, its discourse was solidly grounded in specific problems (e.g., outdated policies, technical challenges) and their potential solutions (e.g., policy reform, technological innovations). But I’m now noticing more conversations about the meta-problem of implementation or, as we call it in the YIMBY (“yes in my backyard”) community, “theories of change.” 

This shift in the conversation is a sign of the progress community’s maturation.

To a first approximation, many of the problems that concern the progress community can be thought of as a simple set: a defined problem plus the solution believed to be the best out of all the options: 

• AI requires more energy. Build nuclear. 
• Science is slow. Change funding to accelerate it. 
• Housing is scarce. Build more housing. 

One step removed, though, is the question of how to implement the given solution. 

By way of metaphor, this is like identifying a physical ailment (the problem), concocting a pharmaceutical intervention (the solution), and then needing to determine the best delivery method for your wonder drug (the meta-problem of implementation). In many cases, the efficacy of the solution is directly tied to the delivery method — imagine a vaccine that does nothing if administered as a pill, but that generates a protective immune response if delivered via an injection.

This shift in the conversation is a sign of the progress community’s maturation: It’s applying meta-science thinking to the social world, and this is where change scales. 

From solutions to implementation

During Progress Conference 2025, I discussed and debated several friction points for implementing solutions with fellow attendees, including the need to identify and inform the right people and institutions, the difficulty of matching a good idea with the right business model, and the challenge of navigating scenarios where someone else’s idea has to lose.  

This conversation is well-trodden ground within YIMBY circles. 

YIMBY Action (where I serve on the board) operates with a people-centric theory of change. As Executive Director Laura Foote puts it, “Politicians don’t listen to polls.” Instead, they overindex on the loudest people in the crowd and the most recent voice they’ve had in their ear. Consequently, we put a lot of effort into enabling YIMBYs to be in rooms with state and local policymakers, preferably armed with megaphones. 

Many of the other policy-oriented organizations in attendance at the conference (including many of the other YIMBYs) have a more elite-driven worldview. Broadly speaking, this means they spend a lot of their time drafting model legislation, writing white papers to get the best ideas into the right ears, and walking the halls of power to help broker deals and cajole elected officials behind closed doors.

The need for political power is just one type of implementation challenge that the progress community faces.

These differences don’t mean our organizations are in stark disagreement with one another — it’s more that we have differences in prioritization. Call it an intellectual division of labor. It’s something we’ve certainly cultivated within YIMBYism and that we should encourage elsewhere. After all, if there’s more than one way to skin a cat, there are probably more than a few paths to bringing back nuclear power. 

Regardless of strategy, YIMBYs operate in a hostile political ecosystem. With respect to the housing crisis, the policy is settled science: to lower housing prices, we must increase supply. Further, everyone knowledgeable on the topic believes that legacy land-use institutions are a binding constraint. YIMBYs still have work to do because a subsection of society extracts benefits from the status quo — when it comes to housing policy, we’re reforming a system that fights back. 

The meta-problem of needing political power to implement policy solutions shapes how the YIMBY community approaches progress. But it is just one type of implementation challenge that the broader progress community faces. 

Criticism through creation

Blake Scholl, CEO of Boom Supersonic, is on a mission to resurrect supersonic flight as a viable commercial enterprise. He was one of the keynote speakers at Progress Conference 2025, and according to his talk, the lack of a viable business model for supersonic flight is the reason we can’t fly from San Francisco to New York in just two hours.

In Scholl’s telling, the original sin was the fact that national governments developed supersonic flight, and not as viable commercial services, but as prestige projects. The Concorde was only in service for a couple of decades, and toward the end of its run, tickets on transatlantic flights could cost tens of thousands of dollars (adjusted for inflation). This kept ticket sales low and contributed to the eventual retirement of the technology from commercial use. 

Scholl thinks his team can succeed where others have failed. His thesis is that by staying focused on developing a product where the economics make sense from the get-go, Boom can start with an up-market product and establish a beachhead. That early success will provide the time and space the company needs to continue iterating on the technology until it is eventually able to provide cheaper supersonic flights to the rest of the market. 

This approach is what I think of as “criticism through creation.” Scholl believes the world could and should have commercial supersonic flight. He has a story explaining why it’s absent: a bad business model led to economically non-viable tech. He also has a solution: create better tech through a better business model. And a theory of change that boils down to: “Fine, I’ll do it myself.”

We could imagine someone with the same priority — resurrect supersonic flight — working on policy to reshape incentives or taking the Nader-esque approach of being a noisy consumer advocate. One issue might be tractable via multiple approaches, and different issues can require fundamentally different forms of intervention — something I was reminded of by Tom Kalil, CEO of Renaissance Philanthropy. 

Spreading good ideas

At Progress Conference 2025, Kalil gave an absolutely illuminating talk, titled “Creating a Marketplace for Outcomes,” during which he detailed all the ways science funding is underoptimized for producing breakthroughs and how we can accelerate science by changing the ways in which we fund research. 

Kalil, who worked on science policy in both the Clinton and Obama administrations, explained the dominant paradigm for scientific funding is currently a federal loan guarantee, which, he pointed out, is only taking on risk against failure and doing nothing to encourage success. He then went through a menu of alternative funding models designed to encourage more productive research, including advance market commitments and incentive prizes. 

Market commitments may sound familiar due to the COVID-19 pandemic, during which governments derisked vaccine development by pre-purchasing a certain number of doses ahead of time. As an example of incentive prizes, Kalil used the DARPA Grand Challenge, which was totally new to me. 

We can use funding to create incentives for researchers and reshape the applied science ecosystem.

In 2004, DARPA (the U.S. Department of Defense’s Defense Advanced Research Projects Agency) created a competition to instigate the development of autonomous vehicles. Entrants would race their autonomous vehicle on a 142-mile course, with the first team to cross the finish line in less than 10 hours receiving a $1 million grand prize. Though the initial challenge was inconclusive (none of the entries successfully navigated the competition course), the winning team from 2005 was promptly hired by Google, which ultimately led to the creation of Waymo. 

The lesson: We can use funding to create incentives for researchers and reshape the applied science ecosystem.

During Q&A, someone in the audience asked Kalil why we don’t see more examples of these approaches if they are so good. His response, which has haunted me, was that these types of programs just aren’t widely known. The challenge now is to broadly disseminate information about them and get the concepts in front of a certain set of federal bureaucrats.

From Kalil’s talk, I got the impression that the intellectual markets in science research are just inefficient. He didn’t name gatekeepers or call out any actors enriching themselves via the status quo — he just noted that the relevant policymakers need to get turned onto some new ideas. This reminded me of the proverbial Chicago School economist seeing a $100 bill on the sidewalk and deciding it must not be real — if it was, someone would have already picked it up.

While my YIMBY mind automatically assumes a hostile political economy, Kalil seemed to be describing a problem that could be solved via a little proselytizing. It was a good reminder for me that, at least for some problems in the world, there may still be meat on the bone of education and thought leadership. 

Nietzsche’s Übermensch vs. Hayek’s entrepreneur 

Throughout the conference, there was a throughline that individual initiative is important — after all, at some point someone has to do something about a problem. Understanding how the progress community thinks about individual action is necessary in order to fully understand how people think about theories of change. 

The term that those in the community use for people who readily take initiative is “high-agency individuals.” To distill the vibe into something tangible: being high-agency is to treat obstacles not as blockers, but as merely preliminary challenges to overcome. 

This can sound a little Nietzschean or even smack of John Galt, which isn’t necessarily off-base. Many of the problems the progress community wants to fix will require overcoming inertia, so a certain force of will and a level of ambition will be required. And yet, I don’t know if that’s the best way to think about agency. 

Let’s reconsider Blake Scholl, the CEO on a mission to resurrect supersonic flight. It would be very easy to interpret him as some heroic Randian industrialist. But after listening to him, what stuck out to me most was his curiosity. 

As Scholl wound his way through explanations of the various ins and outs of present-day aviation, he dropped a real gem of a heuristic: “If you see something that seems dumb, the first thing to ask yourself is what would need to be true about the world to make that thing not actually dumb.” 

Although a bit tongue-in-cheek, I thought it was a really useful way to interrogate the way things work. That bent of mind — that impulse to look at something that doesn’t make sense and ask “Why?” — comes from a place of deep curiosity. It’s much less “dominating will” and much more “entrepreneurial awareness.”

Someone else who really exemplifies this for me is Laura Foote, executive director of YIMBY Action. In talking about the early days of YIMBYism in San Francisco and the founding of YIMBY Action, she once told me that anyone could have done what she did. And really, what it comes down to is that she was the right person, in the right place, at the right time to see the things that other people weren’t seeing — and to act on the things that other people weren’t ready to believe were possible. 

Herein lies what I think is the more powerful version of individual agency. It’s not the superlative genius overcoming all obstacles to lead us into the future — though we should, of course, do our best to keep Harrison Bergeron unshackled — but the person who has cultivated a disposition of restless curiosity that then compels them to take action. 

Connecting the threads

Patience and perspective have to enter the picture here. To stick with YIMBYism as an example, we’ve been working on housing reform for about a decade, and people will sometimes question why we haven’t “fixed” the housing crisis yet. Often implicit in that question is an assumption that the movement has prioritized the wrong things. Obviously, I’m inclined to disagree, but the deeper lesson here is that high leverage doesn’t always mean fast. 

With regard to U.S. housing policy, we live in a different universe from the one we inhabited 10 years ago, but we’re refactoring institutions that developed over the course of a century, so there’s miles left to go. The same could be said of many issue areas that concern members of the progress community.

While we can never convince everyone of everything, building the things we want to build will require convincing some people of some things. Ancient Greek polymath Archimedes once said, “Give me a lever long enough and a place to stand, and I shall move the world.” As a community, I think we are, in many ways, building levers and looking for places to plant our feet.

The progress community still believes in solving specific problems; we’re just becoming more self-aware about implementation as a meta-problem. We’re moving beyond simply identifying root causes to thinking more deeply about how to intervene in systems that are sometimes inefficient and other times actively hostile. Whatever the case, there’s a growing recognition that a solution is worthless without a mechanism for putting it into the world. 

Progress studies as a whole would benefit from a similar meta-conversation focused on theories of change and the practical art of moving ideas from thought experiments to reality. If progress is about problem solving, creating a science of change is just one more problem space to explore on the road to that more prosperous future we all imagine.

This article Rethinking how we think about progress is featured on Big Think.

Walking through Progress Conference 2025 in Berkeley felt like flipping through Leonardo da Vinci’s notebooks: that same dizzying surface area, where a sketch of Florentine waterways sits beside anatomical drawings of human musculature, with mathematical theorems scribbled in the margins. Session topics bounced from longevity biotech to nuclear fusion to maritime transportation to why America hasn’t built its own Shenzhen yet.

The sheer variety was striking, yet everything converged under one tent: progress studies, also known as the progress or abundance movement. For one long weekend at the Lighthaven campus, shipbuilding experts stood next to synthetic biologists next to agricultural reformers, all seemingly unconnected until you realized they share a conviction that material progress has stalled in America, and that restarting it requires cultural transformation and physical building.

Among the crowd, I sensed a yearning for what writer Dan Wang calls “physical dynamism,” the tangible acceleration of the material world that makes tomorrow feel radically different from today. This dynamism has largely vanished from America over the past few decades. But at this conference, I saw glimmers of it returning.

Two projects struck me as particularly emblematic of this resurgence: California Forever and Navier. The former aims to build an entirely new city in southeastern Solano County (between Sacramento and the Bay Area). The latter wants to revolutionize water transportation.

As part of Progress Conference 2025’s agenda, I visited Navier’s assembly plant at its headquarters in Alameda. Water puddles glistened under fluorescent lights. Tattooed engineers worked alongside machinists, assembling carbon-fiber hydrofoils that will literally fly across the San Francisco Bay. An American flag hung from the roof. 

Navier’s founder, Sampriti Bhattacharyya, emigrated from Kolkata, India, to the U.S. and is now building America’s first all-electric hydrofoil boats — if Navier achieves its vision, these vehicles will slash Bay Area commute times from 90 minutes to less than 10.

During the conference, I met another immigrant attempting something even more audacious. Jan Sramek, a 37-year-old former Goldman Sachs trader born in the Czech Republic, is spearheading California Forever, which has spent nearly $1 billion on acquiring land to build its new city. His diagnosis is that America has forgotten how to build cities at scale. His solution? Well, start from a vast amount of farmland and build a mixed-income city with affordable housing, walkable neighborhoods, and integrated public transit — precisely the kind of urban planning that progressive advocates champion, but that NIMBY (“not in my backyard”) homeowners consistently block. It’s an attempt to prove that the “shining city on a hill” is achievable if you’re willing to bypass decades of accumulated regulatory sclerosis.

Immigrants see not only what America could become, but what it risks losing.

Amid all this intellectual diversity, a pattern emerged that the conference rarely named explicitly: The most audacious physical dynamism projects are led by first-generation immigrants. Look closely at the progress movement’s architecture, and you’ll see immigrants everywhere, including at the foundation.

Patrick Collison, the Irish immigrant who co-founded Stripe, not only launched “progress studies” with Tyler Cowen in 2019, but has since become a kind of Medici figure for the movement, funding ventures ranging from the Roots of Progress Institute to frontier research projects that push the meta-lever of progress. 

Heike Larson, who co-founded the Roots of Progress Institute and recruited me as a writing fellow, grew up in divided Germany. During my interview a few months ago, she traced uncanny parallels between Germany’s postwar rebuilding and contemporary China’s development, recognizing in both the psychology of nations determined to prove themselves.

Dan Wang, a Chinese Canadian author of the blockbuster Breakneck, urges American builders to study the advantageous parts of China’s engineering state, not merely as an adversary, but as a mirror and reference point for the state capacity that America has let atrophy.

I write this as a first-generation immigrant from China myself, constantly struck by the sheer ambition of immigrants who come to America to realize their boldest dreams, and by how America has provided that ground for so many years. The founders building most aggressively come from places that followed radically different development paths than America — some spectacularly successful, others failed. They watched Mumbai expand and Shenzhen materialize from fishing villages. They internalized different baselines for what’s possible, and different nightmares of what could go wrong.

That wisdom matters. Immigrants not only bring non-American-centric mindsets and building speed. They carry lived experiences of both progress and collapse, which breeds a particular kind of vigilance. They see not only what America could become, but what it risks losing. Many arrived believing in promises America made to the world, and now they’re trying to hold the country accountable to those promises, to push it to live up to the dream that brought them here. 

If anything, this cosmopolitanism grounded in American reality defines what dynamism actually means to me. 

China as America’s benchmark

Kevin Kelly — one of my favorite writers — recently published his latest book, 2049: The Next 10,000 Days, in Chinese only. In it, he describes “Cool China”: a future version of the nation that produces cool products with a creativity and charm that others will want to imitate. In my observation, today’s China is already serving as a benchmark, foil, and mirror for the U.S. But here’s what strikes me about Progress Conference 2025’s fixation on China: It’s almost entirely abstract. China functions as a building machine in the American imagination, a proof-of-concept that ambitious infrastructure remains possible, while the actual lived reality of Chinese people is invisible to Silicon Valley’s gaze.

The “Cool China” that captivates the progress movement consists of gleaming subway systems, synchronized drone shows, and EV factories churning out vehicles at impossible speeds. The “real China,” I argue, the economic structures governing most people’s lives, the feminist backlash against the state, the slow-motion real estate collapse, the vivid corners of the Chinese internet, barely registers. China becomes a motivational poster for American builders.

And yet, this abstract China is working and helpful. I’m not objecting. I’m marveling at how it’s catalyzing American ambition in ways that feel almost perverse in their effectiveness.

Jan Sramek told me his California Forever project could be America’s response to Shenzhen — just as Shenzhen was originally China’s response to Silicon Valley’s San Jose. It’s a feedback loop of competitive mimicry, each side using the other as proof that their own institutions could be better. During her talk, Navier founder and CEO Sampriti Bhattacharyya invoked the industrial capacity of Chinese shipbuilding, a sector where China now accounts for over 70% of new ship orders globally. 

Dan Wang’s book Breakneck has given people a more grounded context to understand China, as well as the U.S., and has become canonical in the progress community. It reveals new facts about China, but also articulates what American builders already feel viscerally: that the U.S. is trapped in lawyerly procedures while physical dynamism accelerates elsewhere.

Wang’s keynote drove this home with brutal comparisons. California’s high-speed rail has consumed $120 billion since 2008 without connecting San Francisco to Los Angeles — the completion date keeps receding like a mirage. China, starting from scratch the same year, completed its Beijing-Shanghai line in three years for $40 billion. It has now served 1.4 billion riders. The gap is embarrassing.

The audience, engineers, founders, scientists, policymakers, leaned forward as Wang laid out more examples. Many had spent years watching Chinese companies like BYD and Xiaomi iterate on EVs and consumer electronics with a cadence that makes Detroit and Apple look arthritic. Ford’s CEO recently admitted Chinese EVs are not only better but cheaper — a confession that would have been unthinkable a decade ago. During COVID, America couldn’t manufacture N95 masks or test swabs at scale. The defense industrial base sputters: naval ships arrive years late, while drone production stays glacial and expensive.

In another talk, Derek Thompson traced California’s transformation from builder to blocker through two cities. In 1950, Lakewood, once just a cattle ranch, built 17,000 homes in three years, completing one new house every 7.5 minutes. By 1954, it was California’s 19th-largest city. Then came Petaluma’s 1971 growth-control plan, capping new housing at 500 units annually. That model metastasized. Today, Thompson noted, California is far more Petaluma than Lakewood; regulations designed to protect communities now prevent building the infrastructure those communities desperately need.

One of the problems, according to Wang, is a missing link in America’s elite pipeline. The best engineers flow toward $500K AI jobs at startups. The best lawyers flow toward federal clerkships and partnerships. There’s no Tsinghua-to-Politburo equivalent where technical expertise becomes political power. “Yale Law dominates federal power,” Wang noted in his keynote. “For Stanford engineers? Why would they leave Silicon Valley to navigate bureaucracy?”

In Breakneck, Wang writes that he wants Americans to experience what “many Chinese have felt in the past two decades” — the visceral thrill of watching infrastructure materialize before your eyes, of tomorrow feeling radically different from today. But this requires institutional power change. It requires engineers in positions where they can actually authorize building rather than merely advising lawyers who ultimately make decisions.

Expanding American Dynamism

“American Dynamism” emerged as a venture capital thesis around 2022, coined by Andreessen Horowitz partner Katherine Boyle to describe startups tackling hard problems in physical infrastructure, defense, and manufacturing, sectors that Silicon Valley had long ignored in favor of consumer apps. The term metastasized quickly, spreading beyond VC pitch decks into a broader ideological project: a wager that American entrepreneurship could rebuild the country’s atrophied industrial capacity.

But the term itself occupies awkward ideological territory, one I want to challenge and expand. Born from Silicon Valley techno-optimism and cultivated by venture capitalists, it risks becoming another neoliberal fantasy: deregulation dressed as patriotism, celebrating founders while minimizing workers.

America cannot reindustrialize without immigrants and immigration reform.

The immigrant founders I encountered during Progress Conference 2025 complicate and enrich this narrative. Many articulate social democratic visions rooted in cosmopolitan patriotism: infrastructure investment paired with robust labor protections, immigration reform as an innovation necessity, abundant housing and transit as public goods. Their version of American Dynamism is redistributive, physically dynamic, and has strong building capacity.

And I want to articulate a truth the conference barely whispered: America cannot reindustrialize without immigrants and immigration reform. It also cannot do it alone, without borrowing wisdom from countries like China, Korea, Singapore, etc. I envision a different future: one where immigrants and transnational talent, people carrying know-how and the secret formulas across borders, fluent in multiple systems, fuel this dynamism. Radically pluralistic, ambitious, grounded in the lived experience of people who urge America grow into something it’s never quite been.

This matters because one of America’s most enduring myths is that anyone can become American. As historian Oscar Handlin wrote, “Once I thought to write a history of the immigrants in America. Then I discovered that the immigrants were American history.” 

When American identity becomes malleable, absorptive, self-evolving, self-contesting, when it remains a living argument rather than fixed inheritance — its convening power can cut through stagnation. That power depends on keeping the door open, on treating “American” as a verb rather than a noun.

I don’t want to pretend I don’t see the Trump administration’s increasingly draconian immigration policies: the hostile rhetoric, the H-1B visa restrictions weaponized against skilled workers, and ICE raids terrorizing communities. The cruelty is real and accelerating. But what gives me stubborn hope is the resilience both immigrants and the American identity itself have demonstrated. 

Injecting “industrial literacy”

Jason Crawford, founder of the Roots of Progress Institute, closed the conference with a keynote that made the cultural project explicit: a new summer program for high school students, marketed as college prep but really an introduction to “industrial literacy,” a term I found so valuable. It means a visceral understanding of how the modern world actually works, from semiconductor fabrication to power grids to container shipping.

Crawford’s vision draws on economic historian Joel Mokyr’s insight that cultural factors determine whether societies innovate or stagnate. In The Gifts of Athena, Mokyr writes, “What matters clearly is culture and institutions. Culture determines preferences and priorities. All societies must eat — but cultural factors decide whether the best and brightest tinker with machines and chemicals, or perfect swordplay or study the Talmud.”

Crawford wants to shift American culture so that building physical infrastructure becomes as prestigious as writing code or filing briefs. “Students should graduate knowing the hockey-stick curves of human development,” he argued, “but also the vast opportunities still ahead.”

There’s something valuable I found everywhere in the conference, this deliberate cultivation of ambition. Reflecting on the entire Progress Conference 2025 and my experience as a Roots of Progress fellow, I felt wrapped in what I can only describe as examined and chosen optimism: a joyfulness and eagerness about progress that felt both strategic and sincere. This optimism is rare in today’s America, where apocalyptic thinking dominates both left and right, where every technological advance gets framed primarily through its dangers. This is a simple yet encouraging belief that progress is good, necessary, and achievable.

As Kevin Kelly says repeatedly: “Over the long term, the future is decided by optimists.” I believe this deeply. Instilling optimism — even willing yourself toward it. I’ve watched the United Kingdom, paralyzed by apologetic guilt over its industrial and colonial past, unable to build anything ambitious. The constant moral self-flagellation produces neither justice nor dynamism; it left the country stagnant, wrapped in ethical language. The progress community’s refusal to apologize for wanting to build is deeply liberating. Progress is good. We need more of it. The refusal to hedge that claim matters.

This article Physical dynamism and the immigrant’s edge is featured on Big Think.

According to media theorist Marshall McLuhan, every medium is an extension of our physical or mental faculties. The hammer extends our fist, the spear our teeth, the hut our skin, the wheel our feet — and electronic media extend our central nervous system to all of humankind. 

Extended abilities give us new powers but disrupt previous sensory, cognitive, and social settings. As humankind switched to the internet, printing no longer defines the protocols humans live by. Print-based culture is collapsing. 

The era of printing was, in fact, very short by historical standards — about 550 years. Some call this period the Gutenberg Parenthesis — a metaphor popularized by Jeff Jarvis. Within this Parenthesis, communication was centered around fixed, authored, linear, and stable texts. 

Things were different before books. Communication was fluid and situational, passed mostly through direct interaction between people. The internet has largely restored those conditions: digital content is updated, shared, and much of it is user-generated — just like in orality. 

Books withdrew both author and reader from the surrounding reality and turned them inward, into contemplation. Before literacy, it was different — people were immersed in the flow of events. This pattern returns: people are once again immersed in a flow — this time, a digital one. 

As digital media reverse the cognitive detachment typical of print culture and retrieve the situational immersion typical of orality, not only is the Gutenberg Parenthesis closing, but the metaphor of a closing parenthesis — or arc — can be extended to the entire era of literacy. 

The environmental detachment as a media effect appeared before print — it was the effect of writing. Writing separated the known from the knower (Havelock) and made it possible to contemplate things “from outside” interaction with them. 

“Oral Greek did not know what an object of thought was,” wrote Havelock. To catch thought, the flow of words had to be arrested, and words needed to be captured and separated from the thinker. Writing made this possible by putting words before the individual for contemplation. 

Homer, enchanted by the sensory immersion of orality, sang about beautiful wine or beautiful armor. Aristotle, a writing philosopher, was able to discover beauty. No longer satisfied with explanations through sensations and analogies, thinkers began to ponder ideas. 

Detached and introspective thinking, capable of using thought itself as a tool, fully unfolded the effects of writing sometime between the 8th and 3rd centuries BCE — a period that German philosopher Karl Jaspers called the Axial Age. 

According to Jaspers, the Axial Age was a period of human “awakening.” Fundamental philosophical and religious doctrines emerged in China, India, and the Middle East, along with Greece. Humans became aware of their own existence. Historical consciousness emerged. 

Intellectual and spiritual awakening pulled humans out of the natural cycles of thoughtless survival and led them to inquire into the meaning of life, the universe, and everything. The Axial Age began transforming humans’ tribal existence into complex societies. 

“What is new about this age, in all three areas of the world, is that man becomes conscious of Being as a whole, of himself and his limitations. He experiences the terror of the world and his own powerlessness. He asks radical questions,” wrote Jaspers.

Jaspers saw history as a process of emerging human self-consciousness, leading humans to recognize their belonging to humankind. The Axial Age’s spiritual and intellectual power created the unifying force that led humankind to unity (confirmed, paradoxically, by the World Wars). 

Jaspers admitted he couldn’t find a single decisive cause of the Axial Age. From the perspective of media ecology, however, introspective and abstract thinking were the cumulative effects of writing, with its environmental detachment and “inward turn” (Ong). 

In the 21st century, writing is fading. Digital media have twisted the “inward turn” of literacy like a Möbius strip: the user remains immersed in the self physically, but the self is extended outward digitally. 

With this digital twist, cognitive detachment, typical of literacy, is reversing into situational immersion, typical of orality. That is why the historical arc that digital media are closing is larger than just the Gutenberg Parenthesis — it covers the entire era of literacy. 

Electronic and now digital media not only close this historical arc but do so at an incredibly asymmetric speed. “What took several thousand years to complete has taken us only a few decades to reverse: the West now bathes in the emotions of postliteracy,” said McLuhan. 

The Axial Age lasted six centuries and predefined the next two millennia. The new Digital Axis started with the internet, with its most pivotal (so far) events squeezed into the long decade of the 2010s — between the emergence of social media and the emergence of generative AI. 

It was social media that retrieved orality-like immersion, finishing off literacy. Generative AI seems to be the next pivot, beginning something bigger: strong AI (AGI), which will likely pick up the torch of intelligence from its biological predecessor — us. 

The Digital Axis is a long “decade” indeed, yet it’s immeasurably shorter than the Axial Age and the millennia of civilization it inspired. So, we live in the Axial Decade, which is rapidly rewinding the historical Axial Age and reversing its key cognitive and social effects. 

The Axial Age awakened humankind. What will the Axial Decade awaken? Will the awakening of LLMs (or what comes next) lead to the final reversal — also known as the “singularity”? My answer is yes. Jaspers’ “man becomes conscious of Being as a whole” sounds anew in the age of AI. 

It is for this “Being as a whole” that our being transitions from physical to digital. When the user, its medium, and its environment merge into one, this “one” becomes the self-user, reaching true totality and immediacy of (self) immersion. But that’s a story for another time. 

This book focuses on the reversals of the Axial Decade. The turmoil of the Axial Age lasted for centuries and allowed people to adapt through generational change, yet they were stunned. The piled-up disturbances of the Axial Decade are compressed into one generation’s lifespan. 

What are these reversals? According to McLuhan, when a medium or technology is pushed to its extremes or reaches its limits, its full potential, it tends to flip or reverse its characteristics or effects into something different, often opposite.

The clearest example is the effect of cars. Cars enhance mobility and transportation. But when there are too many of them, their effect reverses from speed and freedom to the opposite: traffic jams.

Digital media pushed human interactions to their limits. First, digital speed makes signals move in the networks faster than in our neural system. Never before has the speed of our interactions with the world exceeded our neuro-capacity. This speed pushes us to the limits.

Second, not just communication but nearly all human interactions are digital, and thus they are all reaching ultimate speed. Third, nearly all humankind is digitized, and the demographic transition to digital, which took the long Digital Decade (from “like” to AI), is complete.

So, digital media have 1) sped up our interaction with the world and each other to their limits, 2) encompassed nearly all human activities, and 3) involved nearly all countries and humans. We have reached the limits of speed, scope, and demographics in this new medium.

Reaching the limits of a technology is the prerequisite for reversal. No wonder everything feels upside down. The logic of media is implacable: when digital media reach everyone and everything and speed our interactions up to extreme intensity, things flip into their opposites.

That is why we are living not just through the era of digital transformation but through the era of digital reversals. One tectonic reversal—the Digital Reversal—consists of countless reversals across all aspects of human life.

The abundance of signals reverses into noise. The abundance of facts reverses into fakes. Free access to self-expression on social media leads to abuse by trolls and bad actors and reverses into censorship. The revolt of the public reverses into anarcho-tyranny.

The abundance of choices flips into decision paralysis. The overload of news reverses into news fatigue and news avoidance. The collective mechanism of content selection—the Viral Editor—reverses into collective content policing: the Viral Inquisitor that watches you.

Put to use in digital conversation, text reverses into texting and reverses delayed, solitary reasoning into impulsive exchange. Abstraction reverses into anecdotal evidence. Feelings reverse into intensities. Boosted by empathy overload, identity reverses into credentials.

The business of news media reversed from plentiful ad revenue to desperately seeking digital subscriptions. This business shift reverses news supply into news validation. Journalism reverses into postjournalism, reversing professional standards from impartiality to activism.

Patriarchy reverses into feminism, and feminism reverses into gender fluidity. The camera enabled capturing reality, but eventually reversed its effect into staging reality for the camera to capture. The society of the spectacle reverses into the digital carnival.

Print-based literacy reverses into digital orality. Reading books reverses into asking the search box. Knowledge reverses into knowing. Escalated by digital media, academia flips into activism. Objective truth reverses into crowdsourced importance.

Pushed to extremes by the digital speed of interaction, we experience reversals in all aspects of our existence—politics, culture, epistemology, even sensorium and biology. Will humankind, approaching its limits, itself reverse into something opposite to its biological nature?

Considering AI, that’s now a possibility many ponder. If media extend the user into the environment, then the ultimate extension will lead the user, the medium, and the environment to merge—to become one, just as we were once one with nature. But nature is digital now. 

AI already embodies a part of this merger: AI is a medium that is the environment to itself. What is lacking is only the merger with its user: us—or some new entity, if AI becomes a self-user.

Before it happens, we will enjoy everything reaching its extreme forms and reversing, then reaching new extremes and reversing again. The Digital Reversal may be historically instant compared to Jaspers’ Axial Age, but it fills the entire lives of people living now.

This article The digital age’s reversion to pre-literate communication is featured on Big Think.

In 1924, while imprisoned at Landsberg Prison following the failed Beer Hall Putsch, a 35-year-old political agitator named Adolf Hitler began writing his manifesto, Mein Kampf. In it, he called for the destruction of the Treaty of Versailles, the creation of a new German Reich through territorial expansion, and the removal of Jews from German life. 

Fourteen years later, on September 30, 1938, British Prime Minister Neville Chamberlain returned from Munich to cheering crowds after concluding a meeting with the same Hitler. Waving a signed piece of paper, he assured his audience that “peace for our time” had been secured. Chamberlain, an experienced diplomat, believed he had persuaded Hitler to abandon the very ambitions he had once gone to prison for attempting to achieve. 

In less than a year, Hitler shattered the agreement by invading Poland and launching World War II, putting into action the same agenda he had laid out 14 years earlier. 

This is far from the only example of the West struggling to evaluate threats accurately, with deadly consequences. But by looking at past mistakes — and successes — we can see ways to improve threat assessment in the future.

Pearl Harbor

Roughly two years after Chamberlain’s failed deal, the United States misjudged a threat of its own. 

Washington had intercepted Japanese communications and received warnings that an attack was possible. Most military leaders believed Japan would not dare strike directly at American territory, though. They were confident that the Asian nation would avoid a confrontation sure to bring about its destruction. Yet Japan’s actions told a different story. Its occupation of Indochina, naval buildup, and public rhetoric about a “New Order in East Asia” made clear it was preparing for a direct clash. On December 7, 1941, 353 Japanese aircraft attacked Pearl Harbor, killing 2,403 Americans and crippling much of the Pacific Fleet. 

This prompted the U.S. to enter World War II and look inward to examine its own failures in the lead up to Pearl Harbor. According to the Roberts Commission of 1942, the key lesson was the need to treat enemy intentions and public warnings with far greater seriousness than before. The belief that Japan would not dare carry out its declared intentions left America unprepared. The fatal consequences of that assumption could not be ignored. The tragedy underscored a simple but costly truth: Dismissing explicit threats as bluffs does not make them less real. It only magnifies the price of being wrong.

The Cuban Missile Crisis

In 1962, the Soviet Union began the secret deployment of nuclear missiles to Cuba under the cover of reassuring diplomacy. Soviet officials insisted that only defensive weapons were being delivered, and Foreign Minister Andrei Gromyko personally assured U.S. President John F. Kennedy that Moscow had no intention of placing offensive missiles on the island. 

Yet the calmness of these denials aroused American suspicion. The sudden surge of Soviet freighters, troops, and equipment contrasted so sharply with the peaceful language of Soviet diplomats that Washington ordered intensified U-2 reconnaissance flights. On October 14, those flights revealed missile sites under construction, confirming that Soviet promises had been deliberate deception. Ironically, it was precisely because Moscow never publicly declared its intentions that Washington stayed more alert; unlike with Germany and Japan, the absence of open threats made American leaders scrutinize the gap between Soviet words and actions, allowing them to detect the danger before it was too late. 

The Cuban Missile Crisis suggests that leaders sometimes respond more effectively when adversaries stay quiet and that open declarations of intent, paradoxically, can make governments more likely to dismiss threats as posturing.

Operation Desert Shield

In the years between the Cuban Missile Crisis and the collapse of the Twin Towers, the U.S. confronted a series of crises abroad, from the wars in Korea and Vietnam to hostage takings in Tehran and the Persian Gulf War — all while attempting to uphold a principle of Pax Americana, a term used to describe the period of relative peace and stability in the Western world following the end of World War II, under the military and economic dominance of the U.S. 

In the summer of 1990, Saddam Hussein massed Iraqi forces along the border with Kuwait, a hub of the global energy supply, while U.S. intelligence carefully tracked the buildup. Some regional experts, including Ambassador April Glaspie doubted that Saddam would risk a full-scale invasion. Glaspie later admitted that she and her colleagues “didn’t think he would go that far,” referring to occupying all of Kuwait. But Washington did not dismiss the threat as empty posturing. When Iraqi tanks rolled into Kuwait on August 2, the U.S. responded immediately, launching Operation Desert Shield, deploying forces to Saudi Arabia, and assembling a broad international coalition. 

Washington’s willingness to prepare for the worst made the difference. Unlike at Pearl Harbor, it treated a looming threat as credible even when some experts doubted it. The contrast with earlier failures suggests that success here stemmed from both the hard lessons of the past and the sheer visibility of Saddam’s buildup, which left little room for wishful thinking.

The war on terror

In July 1990, Sheikh Omar Abdel Rahman (“the Blind Sheikh”) arrived in New York having traveled through Saudi Arabia, Pakistan, and Sudan, and quickly became involved in managing al-Qaeda’s financial and organizational network in the U.S. In April 1991, he obtained a green card from an immigration office in Newark, New Jersey. 

Despite earlier U.S. support for the Afghan mujahideen, Abdel Rahman openly denounced the U.S. and issued religious rulings permitting the robbing of banks and the killing of Americans. In one sermon in New York City, he described Americans as “descendants of apes and pigs, feeding off the tables of the Zionists, communists, and colonialists.” Two months later, on February 26, 1993, al-Qaeda detonated a truck bomb beneath the World Trade Center, killing six people and injuring more than a thousand. 

Five years later, in February 1998, Osama bin Laden and Ayman al-Zawahiri issued a fatwa under the banner of the “World Islamic Front,” declaring the duty to kill Americans “in any country, civilian or military.” On August 7 of that year, al-Qaeda carried out near-simultaneous bombings of the U.S. embassies in Nairobi, Kenya, and Dar es Salaam, Tanzania, killing more than 200 people. During interviews in 1998 and 1999, bin Laden promised that U.S. military forces in the Middle East would be attacked, and in October 2000, al-Qaeda operatives rammed a small boat packed with explosives into the USS Cole in Yemen, killing 17 American sailors. 

Later, in the spring and summer of 2001, U.S. intelligence received repeated warnings that bin Laden was determined to strike inside the U.S., culminating in the August 6 Presidential Daily Brief titled “Bin Laden Determined to Strike in US.” Only weeks later, on September 11, al-Qaeda operatives hijacked four passenger planes and carried out the deadliest terrorist attack in modern history, killing nearly 3,000 people. 

This failure was not due to a lack of warnings, but to a deeper human tendency to discount the unthinkable. Leaders and intelligence agencies assumed that al-Qaeda’s threats were posturing, that no group would dare launch such a catastrophic strike against the U.S. The tragedy of 9/11 thus demonstrates more than institutional failure. It reveals how the West has repeatedly had difficulty taking explicit threats at face value, especially when those threats violate its expectations of rational self-preservation.

Modern conflicts

This unwillingness to take threats at face value has led to some of history’s greatest tragedies — at Pearl Harbor and on 9/11, the U.S. had clear indicators of enemy intent and capability, yet the blame afterward focused mainly on failures of coordination.

A similar dynamic has unfolded in recent years with America’s response to Russian aggression. At NATO’s 2008 Bucharest Summit, Vladimir Putin bluntly declared that Ukraine was not a legitimate state. In 2014, he seized Crimea, explicitly tying the annexation to his ambition of restoring Russian control of the region. In 2021, he followed up with a book and speeches outlining his belief that Ukraine should be brought back under Moscow’s authority — by force. 

Western leaders largely dismissed these overt signals, assuming Putin would not be reckless enough to risk crippling sanctions, economic isolation, and a costly war. Even as Russia massed an invasion force of 150,000 troops positioned in Russia, Belarus, and Crimea in early 2022, many European leaders refused to believe an invasion was imminent. As the EU’s top diplomat Josep Borrell later admitted, “We were quite reluctant to believe it,” despite stern warnings from the U.S. government. 

The same blindness appeared yet again on October 7, 2023, when Israel, one of the world’s most advanced intelligence powers, ignored evidence of Hamas’s preparations for a deadly attack. In the months before the attack, Hamas openly trained, broadcast videos, and even had its complete battle plans exposed, yet Israeli intelligence dismissed the possibility of an attack, convinced that Hamas would not be reckless enough to launch a war it was certain to lose and that its exposed preparations were too overt to be taken seriously.

Believing the unthinkable

The surest way to deceive a modern democracy is to show it exactly what you intend to do. Too often, these nations assume such threats are bluffs or exaggerations because their own logic struggles to grasp that someone would be so blatant. In doing so, they add layers of nuance that do not exist and blind themselves to dangers declared from the start, convinced that no one would be so reckless, or so suicidal, as to start a battle both parties know the aggressor cannot win. 

This happened when Chamberlain tried to persuade Hitler, and it happened again when the U.S. dismissed the warnings of a man shouting from a cave in Afghanistan. The West ignores those who spend years openly declaring their intentions, while comforting itself by insisting they do not mean their words. 

Not every threat is imminent. Some potential aggressors, like North Korea, will hint at violence but stop short of openly declaring an intent to strike. Rhetoric becomes dangerous when it is reinforced by consistent ideology, direct public declarations over time, and obvious preparations for action. From manifestos written in prison to the visible movement of troops and weapons toward the battlefield, democracies must be able to identify the signs that mean they need to shift from explanation to preparation: hardening defenses, rallying partners, and above all, being willing to strike first if waiting appears costlier than acting.

This article The West struggles to evaluate threats. Here’s how it can get better. is featured on Big Think.

Tomas Pueyo is the author of Uncharted Territories, a newsletter helping readers understand deeply how the world works today to navigate the world of tomorrow. You can subscribe to it here.

Do you like the Mediterranean? Should we make more seas like it?

Today, the Mediterranean region is one of the best places in the world to live, with an amazing climate, developed economies, and some of the best beaches on Earth — but it wasn’t always like this.

Roughly six million years ago, shifting tectonic plates closed the Strait of Gibraltar — a narrow body of water connecting the Mediterranean Sea to the Atlantic Ocean — turning the area into a desertic wasteland. About 700,000 years later, the strait burst open again, and a brutal megaflood filled the Mediterranean in a matter of months. 

Without that event, we wouldn’t have a Mediterranean Sea — and potentially no Roman Empire, no early European development, no Age of Discovery, no Industrial Revolution, and no United States. The world would be dramatically different.

We don’t need to be satisfied with just one Mediterranean — we have the technology to create more seas like it. So, do we want to create more? And if so, where?

Hunting depressions

The Mediterranean Sea filled with so much water so quickly because its floor is below sea level, which allowed ocean water to rush in once the Strait of Gibraltar reopened. So the first step to figuring out where we could create more seas is to think about where else in the world we have similar depressions.

The Caspian Sea

Did you know that the Caspian Sea is this big? A significant chunk of its northern coast is below sea level, too, and could be filled — but the area is highly populated today, so that’s probably a bad idea.¹

Not too far away, however, there’s a much more promising spot.

The Dead Sea

At its deepest, the Dead Sea — located between Israel, Jordan, and the West Bank — is 2,400 feet (730 meters) below sea level. Its surface, meanwhile, is 1,400 feet (440 meters) below sea level as of 2025, but its water levels are falling fast.

The Dead Sea’s salinity is 34%, nearly 10 times higher than the Mediterranean Sea’s 3.8%. If we created a pipeline from the Mediterranean Sea to the Dead Sea, it would not only add much more water but also make it much less salty. This would allow algae, fish, and birds to return to the sea, perhaps inspiring a name change: the Alive Sea.

The Dead Sea is already home to some resorts, and increasing the amount of water in it while reducing its salinity might also increase tourism in the area. 

At the same time, we could use the water flowing from the Mediterranean to generate hydroelectricity. This could be used to desalinate some of the saltwater and turn the southern Negev Desert into lush agricultural land.

That was the goal of the $10 billion Red Sea–Dead Sea Conveyance (RSDSC), a planned pipeline from the Red Sea to the Dead Sea. Jordan abandoned the project in 2021, though, because it would have required a lot of cooperation with Israel, and Israel’s support was not forthcoming.

Why didn’t Israel cooperate? A number of reasons, including the high cost of the RSDSC and objections from environmentalists. Israel also had a cheaper, faster alternative Dead Sea project under consideration: Mediterranean-Dead Sea Canal (MDSC). 

Alas, Israel has not pulled the trigger there, either, likely due in part to geopolitical instability — the MDSC would require construction in the West Bank. Israel’s water management priorities have also focused on drinking water.

The Qattara Depression

Geopolitics may be too unstable for the MDSC right now, but there is a similar depression that doesn’t have that issue because it’s located entirely in one nation: Egypt.

The Qattara Depression is 300 feet (91 meters) below sea level, and because it’s farther south than the Dead Sea and the Mediterranean, the region is warmer and drier. 

It used to be home to shallow lakes and marshes, but it’s hypothesized that its access to the ocean closed off one day, like what happened to the Mediterranean. Hot, cut off from the ocean, and without enough rains and rivers feeding it, it dried up, leaving behind only sand and salts. 

Where life used to thrive, now there’s just death and desolation — but what if we connected the Qattara Depression to the Mediterranean, flooding it with water?

Parts of the Qattara Depression are just 34 miles (55 kilometers) from the Mediterranean Sea, and we could build pipelines to cover the distance between the two. This flow of water could generate hydroelectricity, too, which could be used to desalinate some of the seawater into freshwater. The additional water and electricity could power new industries, residences, and tourism.

Evaporation from the newly formed Qattara Sea would increase local humidity levels, helping plants take hold and grow, which would reduce nearby erosion. It might also increase rainfall a bit east of the sea.

Creating the Qattara Sea would have a global benefit, too. Because the Depression would hold a lot of water — 290 mi³ (1,200 km³), or about 30% the volume of the Grand Canyon — filling it would reduce worldwide ocean levels by 3 millimeters, and its constant evaporation would slow the rise of ocean levels by up to 6%.

All these benefits come at low cost:

• The limited existing flora and fauna would be replaced by life adapted to the lake. However, there would be more life in general, including fish and birds that are nowhere in the region today.
• The local oases would be disrupted, but the new sea would dramatically increase the economic potential of the region.
• The freshwater table below the region would probably become brackish, but that water is barely used today anyway.

So, why hasn’t the Egyptian government flooded the Qattara Depression? 

Financials. The mountains between the Mediterranean Sea and the Qattara Depression would need to be excavated — an engineering challenge so great that those who looked into it after WWII proposed using nuclear bombs to make the holes² — and a study from 2012 estimated that this would cost nearly $2 billion.³

However, an undersea pipeline that brings water from Turkey to Northern Cyprus — traveling a similar distance (50 miles), and at an incline reaching 80 meters above sea level (the Qattara pipeline would travel down to 250 meters below it) — cost just $430 million. 

That pipeline is only about one-third the size of the one needed for the Qattara Depression, but the order of magnitude is interesting. Also, have you heard of any recent innovation in tunneling costs?

Elon Musk’s Boring Company says it is “targeting an all-in cost of a Loop Transportation tunnel of less than $8M/mile.” The distance needed to tunnel for the Qattara project would be 50 miles, meaning we could get a new sea for just $400 million. 

In 2023, Egypt approved a new feasibility study for filling the Qattara Depression, but while we wait for the nation to decide whether or not to move forward with the project, let’s consider other places we could create new seas — including an outstanding location right in the U.S.

The Salton Sea

The Salton Sea is a massive below-sea-level depression located in Southern California between Palm Springs, Coachella, and the Imperial Valley. 

In the 1950s and ‘60s, it was a rich tourist spot, but today it is a dumpster. What happened?

Throughout history, the Colorado River has alternated between reaching the Sea of Cortez and filling the Salton Sea. The last time the Salton Sea had dried up was around 1700, and it was still a desert in the early 1900s when an accident during the building of the All-American Canal diverted all the water from the Colorado River into the Salton Sink for two years.

For decades, this water made the Salton Sea a tourism destination. It also made the Imperial Valley one of the most productive agricultural areas in the world because farmers could flood their farms with the river’s sediment-rich waters. These, in turn, would flow into the Salton Sea, refilling it.

But as water became scarce in the American Southwest, the allocation to Imperial Valley farmers was reduced, overflow stopped, and the Salton Sea started drying up. Fish started dying off, birds stopped coming, and tourists abandoned the area. As the waters receded, the winds started blowing the sand of the exposed lakebed into the lungs of people in the area. 

The Salton Sea is now the biggest man-made source of hazardous dust in the U.S., at levels 10 times higher than the acceptable standard. Los Angeles has spent more than $1.2 billion trying to suppress the dust, pouring 30 billion gallons of water onto the lakebed each year, but air-quality problems remain. 

How do you solve this?

Make the Salton Sea great again

The Salton Sea is 230 feet below sea level, and it’s just 120 miles from the Sea of Cortez, with no mountains in between, so digging a simple canal would suffice to bring ocean water to the Salton Sea, reducing its salinity in the process.⁴ It would cost just $800 million and $40 million in annual maintenance. Adding a concrete lining would bring the costs to a bit over $2 billion.

For that cost, we would get several benefits:

• The Salton Sea tourism industry would return.
• Seawater contains plenty of lithium, which we need for batteries and other technologies. This lithium could be harvested from the sea.
• We’d revitalize the area’s real estate. This alone would create trillions in value.

Additionally, the area has plenty of desert where we could put solar panels. We could then use the electricity to desalinate water for drinking and industrial use at $0.30/ton — that’s much cheaper than what that water costs in California today.

That desalinated water is more expensive than what agricultural water costs today — Imperial Valley farmers pay ~$0.015/ton — but it could still be used for some agricultural products. The difference would add just $0.07 per pound of tomatoes, for example.

Casey Handmer, the founder of Terraform Industries, has a similar plan, but he proposes refilling the Salton Sea with desalinated water instead of saltwater. The obvious drawback is that it would cost tens of billions of dollars. But there’s a little-known fact in favor: Somebody else is already refilling an inland sea with desalinated water.

The Sea of Galilee

Virtually all of Israel’s drinking water is desalinated. If you consider all uses (drinking, industrial, and agricultural), over half of its water is desalinated. The Israelis have achieved that by establishing six major desalination plants. 

In periods of drought, though, Israel draws water from the Sea of Galilee to the north. Its water levels have been falling in recent years, so in 2022, Israel launched a project to begin refilling it with desalinated water. That refilling will begin in late 2025. 

If Israel can afford it, California’s economy — which is eight times bigger — can, too.

More lakes in the U.S.

We wouldn’t have to stop at reviving the Salton Sea. We could create several new lakes in the U.S. — and each time you have a lake, you can build a new community. 

As its name illustrates, Salt Lake City exists because of the Great Salt Lake. Nearby Provo exists because of Utah Lake. Tahoe is an immensely valuable lake in the middle of the Californian Sierra Nevada. Las Vegas is an oasis in the middle of the desert that would not have been possible without Lake Mead, which was created by the Hoover Dam. Just south of that, Lake Havasu birthed Lake Havasu City, a community of 50,000 people.

A few existing lakes in the U.S. Sierra Nevada and Rockies could be revived to anchor new communities, but what if we created brand-new lakes? There were many more lakes in the U.S. just a few thousands of years ago that we could use as a basis for these.

We could revive any of these lakes if we just sent enough water to them. But from where? Most of these places are more than 3,000 feet above sea level. Pumping water up to them from the ocean would be very expensive, so where else could we get the water from?

The Colorado River. 

The vast majority of its waters go to low-lying areas of California and Mexico. But what if we desalinated seawater for these regions instead and kept the Colorado River’s freshwater in Nevada?

Then we would barely need to do any pumping. The only cost would be desalination, which at $0.30/ton should be quite affordable. All that water could be diverted into recreating the lakes that used to exist. And for every new lake, we could create a new community.

With its own rules.

With its own culture.

With its own urban design.

With its own priorities.

For centuries, humans saw nature as what it was to them: brutal, cruel, something to conquer to eliminate pain. But when we did, we overplayed our power, leading to the destruction of nature. Now, the pendulum has swayed in the opposite direction: We fear touching anything. Some even advocate for a world with fewer humans. 

But that’s not the right lesson. Of course we should be thoughtful with megaprojects. We should weigh the pros and cons and try to mitigate the cons. But that shouldn’t freeze us into inaction. 

We should be modifying the world. We should make it a better place. We should flood the Qattara Sea. We should refill the Salton Sea. We should create new lakes in the Rockies. We should engineer the Earth again to make it a better place for humans and wildlife.

1. Russia could, however, refill it with water from the Black Sea, as today Russia is diverting so much water from the Caspian that its levels are going down fast.
2. There were other potential concerns, such as coastal erosion or the risk of triggering unexploded mines from WWII, but these seem tiny compared to the excavation issue.
3. “Integrated Wind and Solar Qattara Depression Project with Pumped Storage as Part of Desertec” Weisensee & Ragheb (2012)
4. The Salton Sea’s salinity is 4.5-5%, so about 50% higher than the Pacific Ocean’s.

Tomas Pueyo is the author of Uncharted Territories, a newsletter helping readers understand deeply how the world works today to navigate the world of tomorrow. Subscribe to it here.

This article Seaflooding: How we could engineer the next Mediterranean is featured on Big Think.

Viewed from orbit, Jackass Flats — situated in southern Nevada about 65 miles northwest of Las Vegas — could easily be confused for Mars. The alluvial basin is full of tan and gray regolith, hued slightly red, and almost completely surrounded by carved, rocky hills. It was here, a half-century ago, that NASA engineers tested nuclear rockets intended to get us to the Red Planet by 1978.

Officials had even grander hopes for the descendants of those rockets. They were planned to be mules for a permanent lunar base by 1981, propulsion systems for deep space probes to Jupiter, Saturn, and the outer planets, and engines for “space tugs” and shuttles ferrying payloads and people from low Earth orbit (LEO) to space stations around the Earth and the Moon. NASA even envisioned a “Grand Tour” of the Solar System propelled by nuclear rockets, taking advantage of a planetary alignment that happens every 174 years to visit Jupiter, Saturn, Uranus, and Neptune with one sweeping mission between 1976 and 1980.

Why didn’t this glorious space future come to pass? What went wrong? Well, nothing. Testing these rockets went astoundingly well, in fact.

“The Rover/NERVA program conducted through the decade of the 1960’s was a highly successful technology program,” a NASA contractor report praised in 1991. “Its goals and objectives were to demonstrate the feasibility of a nuclear rocket engine system for space application. This ‘Proof of Concept’ program was mission oriented and culminated in the successful demonstration of a ground test engine system.”

Nuclear rockets were nearly ready to launch into Earth orbit for the next phase of testing. Then everything changed.

Why don’t we have nuclear propulsion now?

The short story: America won the lunar space race. Afterward, public attention soon turned back to earthly affairs: ending the Vietnam War, record inflation, an oil crisis. Politicians preferred cost-cutting to pricey space adventures. In January 1973, the Nixon Administration abruptly canceled NERVA, the Nuclear Engine for Rocket Vehicle Application program, after 17 years of research and development and roughly $10 billion spent in 2025 dollars. The Space Shuttle program became NASA’s primary focus. It turned out to be a successful, but decidedly more local, venture.

Nuclear propulsion received piecemeal federal funding and attention over the ensuing decades, but nothing close to what it enjoyed in the 1960s. There was Project Timberwind in the 1980s, the Space Nuclear Thermal Propulsion program in the 1980s and 1990s, Project Prometheus between 2003 and 2006, and most recently, the DRACO program between 2023 and May of this year. All projects were ultimately canceled.

Bhavya Lal — a Professor of Policy Analysis with RAND School of Public Policy and a former NASA associate administrator for technology, policy, and strategy — recently co-authored a report that tallied government spending on nuclear propulsion and explored why the programs failed.

“The US didn’t fail to deploy space nuclear systems because we lacked the physics or the funding or the people,” she said at a Washington Space Business Roundtable event in July. “What we didn’t have was mission pull, institutional coherence, and a sense of scale.”

In other words, every single government program focused on developing nuclear propulsion has been a side project. These missions are susceptible to cuts when politicians look to reduce spending. But despite decades of false starts, the case for nuclear remains strong.

The case for going nuclear

Nuclear space propulsion comes in two flavors: nuclear thermal rockets and nuclear electric propulsion (NEP). In nuclear thermal rockets, a small nuclear reactor heats up a propellant — often liquid hydrogen — to blazing temperatures in the thousands of degrees. Then, just like a conventional chemical rocket, this propellant is transformed into a hot gas and expelled out of a nozzle, blasting the rocket forward. Nuclear thermal rockets are roughly twice as fuel-efficient as chemical rockets while also packing a decent amount of thrust. Not as much as chemical rockets, but more than enough to propel a spacecraft out into the Solar System once a vehicle is in space.

NEP, on the other hand, utilizes a nuclear reactor to electrify thrusters that provide propulsive force. These thrusters put out minuscule amounts of thrust compared to chemical or nuclear thermal rockets, but they are up to 20 times more efficient than conventional rockets. This means they can accelerate spacecraft for much longer durations, trading slower starts for greater efficiency and range, making transit times comparable to other options. A spacecraft equipped with nuclear electric propulsion could also utilize its nuclear reactor to power the entire mission. In space, where mass-efficiency is everything, this could be a game-changer. Bhavya Lal, a Professor of Policy Analysis with RAND School of Public Policy and a former NASA associate administrator for technology, policy, and strategy, explained in an interview with SpaceNews.

“The energy density is so high, you can have a Mars mission with uranium the size of a marble,” she said. “But beyond that, there is no way to do a whole lot of science without nuclear. A few years ago, we went to Pluto with the [New Horizons] mission. The probes whisked past Pluto. Why? Because we didn’t have the delta V to go into orbit… All we have is basically 24 hours worth of data. So we just get so much more value for our money if we have nuclear.”

Almost all spacecraft sent to deep space — including the Voyager probes, Juno, Cassini-Huygens, and Galileo — must execute gravity assist maneuvers, slingshotting around planets to gain velocity for their vast treks. These circuitous routes greatly lengthen mission durations and limit launches to tight windows. Nuclear propulsion can free spacecraft from these constraints.

A NASA all-in on nuclear

So why not make nuclear propulsion a central focus at NASA rather than a side project? Before his nomination to lead the space agency was unceremoniously axed earlier this year as collateral damage of President Donald Trump’s break-up with Elon Musk, Jared Isaacman intended to do exactly that. The entrepreneur, pilot, and commercial astronaut, who was widely viewed as an excellent choice to lead NASA, planned to make nuclear electric propulsion one of his signature issues.

“I think the best way to boost morale and inspire the next generation is to ‘get the big missions done!’: Return American astronauts to the Moon, establish an architecture for routine and affordable missions, and lay out a credible path to Mars,” Isaacman told Big Think in an email. “NEP has an important role to play in that. It also properly calibrates NASA to focus on the “near-impossible”—the kind of work that other agencies or companies aren’t capable of doing or don’t have a business case to underwrite.”

In an op-ed published in August in RealClearScience, Isaacman teamed up with 2012 presidential candidate and former Speaker of the House Newt Gingrich to argue that NASA should build a fleet of NEP-powered spaceships, writing that it is “something no other agency, organization, or company is capable of accomplishing.”

“If America wants to lead, NASA must take on the hard problems again and do the near-impossible. It must urgently deliver the systems only it can build—leaving routine operations such as Earth-to-orbit delivery to the healthy commercial launch industry,” Isaacman and Gingrich wrote.

To make budgetary room for this bold endeavor, the duo advocated cutting the behind-schedule and over-budget Space Launch System (SLS), the super-heavy lift rocket intended to take astronauts back to the Moon with the Artemis program. After 14 years of development and roughly $30 billion spent, the rocket has had just one successful launch, in 2022. Isaacman clarified to Big Think that NASA should only pivot from SLS after the initial lunar objectives have been met.

“At some point we will need to move on from a $4 billion disposable rocket to something more sustainable for frequent lunar missions.”

In an email interview, Gingrich did not mince words in expressing his negative view of SLS.

“The Space Launch System is hopeless but survives by porkbarrel pressure,” he said.

Opening up the Solar System

NASA’s pioneering work in chemical rocketry paved the road for the now-thriving private industry in the U.S., opening up Earth orbit for business. A new focus on nuclear propulsion could open up the Solar System itself.

“The commercial launch industry is healthy and will continue driving down the cost to put mass in LEO,” Isaacman said. “We can then leverage NEP to efficiently transport mass to the Moon, Mars, and beyond. Think of it like a transcontinental railroad – not the fastest way to move a lot of mass, but certainly the most efficient.” 

“In the future, NEP could be the enabling technology for keeping cryogenic propellants cool on deep space missions,” he added, “providing power in regions where solar is less useful, enabling surface propellant manufacturing on Mars for return missions, and even supporting commercial applications like asteroid mining.”

For Gingrich, this nuclear propulsion is not just aspirational; it is an existential necessity. For well over a half-century, it’s been a given that the U.S. and democracy dominate space. A growing number of experts are now warning that China and authoritarianism could take over that leadership within the next decade. 

“We are on the edge of having freedom or communist totalitarianism lead the way into space,” Gingrich told Big Think. “It is historically vital that freedom leads the way and defines the rules.”

According to recent reports, Trump may be reconsidering his decision to rescind Isaacman’s nomination to lead NASA, so a huge nuclear propulsion advocate may soon lead the world’s premier space agency.

Isaacman told Big Think that he wants to see the routine at NASA shaken up.

“The status quo at the agency is not working — that is evident in every Inspector General report. This is not the time for stagnation as China looks to overtake America’s lead in the high ground of space. We need to challenge the status quo, recalibrate around NASA’s true mission and get back to making history. The country needs NASA and the world needs NASA. And I want to see the agency return to delivering world-changing discoveries.”

This article Why NASA should go all-in on nuclear propulsion is featured on Big Think.

Is AI in a bubble?

That’s the basic yet seismic question on a lot of people’s minds. But here’s the thing: It’s oversimplified, attempting to color an unprecedentedly gray moment either black or white. And what does the query even mean? If you’re asking about whether or not the valuations of certain AI startups and the companies that supply them are overvalued relative to their current financials, there’s a strong case for answering in the affirmative. If you’re asking whether the hype over AI has raced ahead of the technological landscape in regards to it attaining artificial general intelligence or rapidly destroying the labor market — the answer might be “probably.”

But if you’re asking whether AI will ultimately fizzle out and go down in history as the fever dream of a science-fiction-obsessed Silicon Valley cult, willed into existence by billions of FOMO-fueled venture capital dollars, the answer is undeniably “no.”

Where is AI adoption, anyway?

Perhaps the best data available to help us answer any form of the omnipresent “bubble” question is AI adoption rates. Are individuals and businesses actually using AI? Are they deriving benefits? Are they integrating it into what they do each day? 

Let’s look at businesses first. QuantumBlack, McKinsey’s AI arm, published a report in June showing that 8 in 10 companies use generative AI — so the technology seems to be catching on pretty quickly.

However, an MIT NANDA report made public two months later appeared to reaffirm skeptics’ canary calls. The researchers analyzed 300 publicly disclosed AI initiatives at various companies, interviewed representatives from 52 organizations, and surveyed 153 senior leaders at major industry conferences.

“Despite $30-40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return,” the authors reported.

That headline-friendly finding spread quickly, trumpeted by a range of media sources as evidence that the presumed AI bubble is about to burst. The previously published QuantumBlack report also indicated that AI had a similarly lackluster effect on companies’ bottom lines, though it failed to generate buzz at the time.

Moreover, starting in June, companies of all sizes reduced their use of AI, according to the U.S. Census Bureau.

Both the MIT researchers and QuantumBlack analysts would say that this is a temporary blip, however. Given how the media widely covered their reports, one might think they hold negative views of AI’s practicality in the business world, but the opposite is true. In reality, both teams gushed that AI is already transforming the way people work and is set to overhaul how businesses operate. 

“For the first time in human history, you can manipulate technology with human language, not a computer science language,” Alexander Sukharevsky, a Senior Partner at McKinsey and the global leader of QuantumBlack, told Big Think in an interview. “It’s always hard to predict the future,” he cautioned, adding: “If I look at the current adoption and willingness of organizations to go for a transformation, it’s the highest ever in my career.”

Two trends help fuel Sukharevsky’s and the MIT team’s optimism over AI. The first is that while most enterprises haven’t derived tangible advantages from adopting AI just yet, individuals have.

An AI “shadow economy” at the workplace

“As of late 2024, nearly 40 percent of the U.S. population aged 18-64 uses generative AI,” a trio of economists reported earlier this year. “Twenty-three percent of employed respondents had used generative AI for work at least once in the previous week, and 9 percent used it every work day.”

If those statistics don’t seem groundbreaking at face value, consider them in a historical context.

“Relative to each technology’s first mass-market product launch, work adoption of generative AI has been as fast as the personal computer (PC), and overall adoption has been faster than either PCs or the internet,” the authors noted. Granted, PCs and the internet were harder to initially embrace due to cost and difficulty of setup, but AI adoption is right in line with two pivotal products of our current technological era, evincing its long-term transformative potential.

The MIT team witnessed this firsthand while researching their report. Though AI initiatives floundered at the business level, AI itself was widely used among the workforce.

“AI is already transforming work, just not through official channels. Our research uncovered a thriving ‘shadow AI economy’ where employees use personal ChatGPT accounts, Claude subscriptions, and other consumer tools to automate significant portions of their jobs, often without IT knowledge or approval.

The scale is remarkable. While only 40% of companies say they purchased an official LLM subscription, workers from over 90% of the companies we surveyed reported regular use of personal AI tools for work tasks. In fact, almost every single person used an LLM in some form for their work.”

These employees had a solid grasp of what it would take for AI to succeed at a grander, enterprise-wide scale: learning and memory. AI must be able to retain information over extended periods of time and adapt to changing circumstances. Which brings us to the second reason that both the MIT and QuantumBlack teams are certain of AI’s integral future in business despite its tepid success so far: AI systems with those abilities are here and are now starting to be rolled out.

Unleash the agents

The overwhelming majority of AI systems used in businesses today are easy-to-implement, shareholder-friendly chatbots. But while these might slightly improve productivity at the worker level — summarizing meetings, making images, or writing emails — they aren’t going to deliver the revolutionary change to workflows that drives major returns on investment.

Chatbots are primarily conversational tools designed to answer questions from a predefined knowledge base. AI agents, on the other hand, are sophisticated, autonomous systems that can analyze information, make decisions, and take actions to achieve goals. Fundamentally, chatbots are reactive, whereas AI agents are proactive. A chatbot is like a calculator — essentially a tool. An agent is a collaborator with a calculator of its own.

“Think about an infinite army of interns that is able to do a lot of simple tasks,” Sukharevsky said of agentic AI. “But you need to orchestrate them, you need to upskill them, you need to provide them with the right information.

Agents are in the early stages of implementation, and these have the potential to deliver on the grander promise of AI.

“Don’t expect the first time to see a masterpiece,” Sukharevsky said. “But it actually evolves. What we see today is the worst … If you look at the technology a year ago, two years ago, and if you look at the technology today, or twelve months down the road, you get completely different outcomes.”

Agentic AI’s rapid progression evinces its shift toward widespread, practical use in enterprises. The MIT researchers found more evidence for this trend.

“In the next few quarters, several enterprises will lock in vendor relationships that will be nearly impossible to unwind,” they wrote.

The consulting firm Source echoed what the researchers were seeing in a trend report published in mid-September. Some 55% of its clients planned to invest in organizational restructuring during the next 18 months, almost entirely due to AI.

“It is impossible to hide from the impact of AI,” the authors wrote. “Few organisations—if any—do not have a roadmap for AI implementation.”

What might agentic AI look like when integrated into a business? QuantumBlack analysts provided a few examples. In e-commerce, agents could observe a user’s behavior, shopping cart content, and purchase history to offer product recommendations in real time. In supply chain management, agents synced to internal and external data sources could continuously forecast demand, allocate stock, determine optimal transport, and more. At a retail bank, agents could formulate credit-risk memos on prospective clients, a task that can currently take a human worker two to four days to complete. Instead, that person can check over the AI’s work and then make the loan decision.

What about us?

If AI agents are essentially digital workers, and you can create a ton of them, where does that leave human workers? 

The broader topic of job loss in the era of AI is widely covered, heavily debated, and nearly impossible to predict in the long term. The only thing we can say for sure is that the nature of work is going to change.

The MIT researchers did hear about some workforce effects directly from company leaders.

“Organizations that have crossed the GenAI Divide are beginning to see selective workforce impacts in customer support, software engineering, and administrative functions … Executives were hesitant to reveal the scope of layoffs due to AI but it was between 5-20% of customer support operations and administrative processing work in these companies.”

The researchers also asked workers in these companies for their opinions.

“Concerns about workforce impact were far less common than anticipated. Most users welcomed automation, especially for tedious, manual tasks, as long as data remained secure and outcomes were measurable.”

Sukharevsky briefly described the role of a human worker in an agentic AI future.

“Your role is to find the right team members, the right skills, the right data, and to put them against the right mission. If you invest enough time, that [AI] intern all of a sudden becomes an expert one day — your peer — and it can add a lot more value.”

Effects of the continuing AI rollout will not be felt equally in all sectors, the MIT team learned. In healthcare, energy, and advanced industry, executives didn’t anticipate any hiring reductions over the next five years. In technology and media, however — where AI is coding, producing images, writing scripts, and crafting video — staff reductions are underway now and more are expected in the near future.

But new jobs and industries will spring up thanks to agents, Sukharevsky said. He’s already seeing it.

“Newcomers are creating completely new business models that you couldn’t imagine in the past, and you couldn’t imagine them because the unit economics couldn’t fly, now they finally fly.”

Sukharevsky stressed that it will take years before agentic AI is incorporated into large, established businesses. Legacy software architecture, human perception, management, and governance are all barriers to successful adoption. 

“You’re seeing mixed results because you need to go ‘all-in’ to get the outcome. It’s difficult, it’s expensive, and it’s not immediate.”

But it is definitely coming. QuantumBlack offers some prebuilt agents and a builder software product for companies to create their own.

“What we’re trying to do is close the gap between the promise of AI and the reality,” Sukharevsky said. “We already have a few dozen agentic transformations under the belt and ongoing, so this is real.”

The answers we seek

Humans generally aren’t hardwired for patience. When faced with uncertain circumstances, we grasp for certainty. After the MIT team published their report, many onlookers grasped onto the notion that AI wasn’t yielding financial results; thus the anxiety-inducing structural change to work and life that AI represents might never materialize. But delving deeper into the report and reading others like it makes clear that AI’s broad effects are only beginning to manifest. 

Change is undeniably afoot. AI’s rise is often compared to the Industrial Revolution. Whether AI will actually result in comparable societal upheaval and economic reorganization is anybody’s guess. The only certainty is that we won’t get the answers we seek in a single report, from the latest adoption data, or even over a matter of months or years. The Industrial Revolution played out over eight decades. Even if the AI economic transformation plays out four times faster, we still have a long ride to go.

This article AI adoption rates look weak — but the data hides a bigger story is featured on Big Think.

We live in a golden age of social trend-spotting. Have you heard of the crisis of democracy, the loneliness epidemic, or the decline of empathy? Public intellectuals, journalists, and politicians often cite broad social trends like these to explain the current state of the world, but many of these trends turn out to be illusory.

Take, for example, “crisis of democracy,” the idea that since the early 2010s, the world has experienced democratic backsliding, with countries becoming more authoritarian and repressive. A landmark article published last year by scholars Andrew Little and Rachel Meng shook the field of political science by demonstrating that most evidence for these claims comes from “democracy scores” based on the subjective opinions of expert coders toward, for example, whether a country’s elections are free and fair. When examining more objective measures of democracy, such as electoral competitiveness and constraints placed on leaders (term limits, for example), Little and Meng found that global trends in democracy are largely flat — some high-profile countries, such as Venezuela, have experienced backsliding, while many others, such as Tunisia, Niger, and Madagascar, have experienced increases in democracy. 

Similar measurement issues surround claims of a “loneliness epidemic,” the idea highlighted by medical experts, ranging from the World Health Organization to former U.S. Surgeon General Vivek Murthy, that people are becoming lonelier. Again, the data underlying this claim are mixed at best, as journalist Faith Hill notes in an article from The Atlantic titled, “The Myth of a Loneliness Epidemic.” Hill cites a 2022 meta-analysis of loneliness studies around the globe, which states, “Findings are inconsistent and therefore do not support sweeping claims of a global loneliness epidemic.” A global analysis of loneliness trends among adolescents, a group considered high risk for loneliness, revealed that from 2003 to 2017, adolescents’ self-reported loneliness increased in a few countries (like Namibia) and decreased in a few countries (like Seychelles), and most countries showed little change in loneliness over time. 

Some trends are also simply prone to reversing. For example, the phenomenon termed the decline of empathy (also known as the “empathy deficit”) refers to a much-cited 2011 study led by psychologist Sara Konrath, demonstrating that from 1979 to 2009, American college students’ self-reported empathy dropped. Yet, an updated analysis of more recent data by Konrath and colleagues showed that empathy has rebounded since 2009. As Jamil Zaki wrote for Vox, other measures of empathy, such as volunteering and donating to charity, are trending upward. Despite this evidence, a recent report from the Muhammad Ali Center reveals that most Americans believe that empathy has declined over the past four years.

Even trends with robust empirical backing are more complicated than they seem at first glance. Consider the “youth mental health crisis,” which documents real trends in rising depressive symptoms and anxiety around the globe over the past 15 years. Yet, the 2024 World Happiness Report, comprised of extensive global data from Gallup, documents that “Overall, globally, young people aged 15-24 experienced improved life-satisfaction between 2006 and 2019, and stable life satisfaction since then.” For all of the discussion of rising psychological distress among adolescents, there is relatively little discussion of rising happiness. 

Due to variations in when data are collected, where they are collected, and the instruments used to collect them, many trends that receive extensive media coverage are not as straightforward as they seem. So why do they have such staying power in our minds and in popular discussion? 

Grasping at patterns

My informed guess comes down to a single word: pareidolia, defined as “the tendency to impose a meaningful interpretation on a nebulous stimulus.” If you’ve ever looked up at the clouds and spotted recognizable shapes — a face, the outline of an animal — that’s pareidolia. As a social psychologist, much of my research has examined how people impose meaning on things that feel difficult to understand. Classifying something as a trend, even if it’s not, can provide comfort when the world feels chaotic. Grasping at patterns in data helps us make sense of the daily cascade of information from news alerts about wars and plane crashes to reports of advances in artificial intelligence and groundbreaking medical treatments. Never before have we encountered such a diversity of information, people, and opinions at such a rapid pace. In fact, if there is any trend that I believe is truly robust, it is simply that the world has become more complex. 

A 2024 study of over 400,000 survey respondents answering questions about social, political, and economic values by psychologists Joshua Jackson and Danila Medvedev confirms this complexity. Their analysis found that across 76 countries, between 1981 and 2022, people’s values have diverged wildly across countries on issues ranging from abortion to immigration to the importance of obedience in children to how justifiable it is to accept a bribe. The world is not only changing, but changing in non-uniform ways. 

Humans are notoriously bad at processing this type of nonlinear information, and to grapple with such complexity, we seek out and perceive linearity, even when doing so is misguided. Research shows, for example, that people mistakenly believe the U.S. has made linear progress on racial and gender equity, and that climate change concerns have increased over time, even when these phenomena are largely nonlinear. People also mistakenly believe that the population’s proclivity for kindness and honesty has decreased linearly over time, even though these aspects of morality have barely budged since scientists began measuring them. In fact, society changes far more slowly than people perceive, and over periods of decades, many social and psychological phenomena simply remain flat. 

Perceiving social trends may provide meaning amid the world’s complexity, but in seeking to make sense of what feels like unprecedented times, it might be just as comforting, and certainly more accurate, to remember that much of what we are experiencing now is quite precedented.

This article Most social trends aren’t what they seem is featured on Big Think.

Over the past three years, large language models like ChatGPT have gone from curiosities to everyday tools used on college campuses around the world. While some professors embrace them, others ban them. Many institutions fall somewhere in between, setting vague policies or relying on detection tools that have been proven ineffective in many instances. AI may be the most significant innovation in education since the personal computer, and although this revolution has many different characteristics, it echoes past battles over calculators and computers — battles that were ultimately lost by those who tried to resist.

Technology doesn’t wait for policy, and as a current undergraduate student, I believe that the sooner schools catch up, the better we can use these tools to improve learning rather than undermine it. Still, an important question remains: Is it fair to compare AI with past innovations like calculators and the early internet, or is this a fundamentally different challenge?

AI is not the first technology to disrupt higher education. In the 1970s, the pocket calculator triggered a wave of backlash among educational institutions. Teachers warned that it would weaken students’ arithmetic skills, and some schools tried to ban calculators altogether. But others saw the potential: If students no longer had to do long division by hand, they could focus on bigger-picture math problems. Eventually, calculators became standard classroom tools, allowing students to shift their focus from manual computation to understanding formulas and solving higher-level, conceptual problems. Studies show that calculators can improve conceptual understanding when used correctly.

This same cycle repeated in the 1990s with personal computers and the early internet. Critics feared that spell-check and copy-paste would erode writing skills, and that search engines like Google and communal encyclopedias like Wikipedia would replace real research. And yes, some students misused those tools. But once schools embraced the technology and taught students how to use it well, evaluate sources, and cite correctly, their academic work improved. Students were no longer limited to the outdated books in their campus libraries, but suddenly had access to a multitude of books, articles, and datasets in multiple languages, at any time.

The cycle of resistance and delayed acceptance is a recurring phenomenon in large institutions, especially those with long-standing traditions in education, such as Columbia University. These universities, responsible for the education of millions of Americans, cannot afford to change course without serious caution. Even when faculty are eager to adapt, such as by updating policies on AI use in student essays, their efforts are often delayed by the university’s complex bureaucracy and layered approval processes. These systems are designed to ensure thoughtful decision-making, but they can struggle to keep pace with rapid technological change. For example, a 2024 global survey conducted by the Digital Education Council found that 86% of students already use AI in their studies, underscoring the technology’s rapid and widespread adoption across disciplines. 

However, it’s clear that the AI revolution is broader and more complex than past technological shifts. Instead of simply speeding up our work, AI can perform tasks that once required deep thinking and creativity, such as writing code or entire essays. Steve Jobs famously referred to the personal computer as a “bicycle for the mind” in 1981, believing it could enhance human intelligence, especially in education, the area where he envisioned the personal computer having the most impact.

The debate about integrating AI into the education system mirrors earlier debates, but it feels louder and more urgent. ChatGPT can help students draft essays, debug code, explain complex concepts, or practice new languages. Its capabilities dwarf those ushered in by calculators and the internet. For this reason, it can easily be misused, but banning it outright, as many universities have attempted, is a battle that was lost before it even started. Telling students not to use a tool that is nearly undetectable and freely available won’t stop its use; it will only push it underground and widen the gap between students who are proficient in using it and those who don’t yet know how to use it effectively.

Moreover, the AI detectors that many universities rely on as a first line of defense have proven deeply flawed; for example, some have flagged writing from international students because their sentence structure tends to be simpler. The current tension reveals a deeper problem. An experienced English professor probably doesn’t need software to spot AI-generated essays: The tone, structure, and sudden leap in fluency are often glaring to a trained reader. But without empirical proof, there is no ethical way to penalize the student. Intuition, no matter how informed, cannot serve as formal evidence. This leaves educators in an impossible position; they can either ignore the changes they notice or act on suspicion using imperfect AI detectors. 

At the same time, my anecdotal experience suggests a strange double standard is emerging. In one of my classes, for example, the professor explicitly banned the use of AI but told us the assignment would be made harder because he assumed we’d use it anyway. On the other hand, some students who are unfamiliar with AI or choose not to use it are falling behind because the expectations for writing and coding have quietly shifted. Rather than fostering an environment of uncertainty and mistrust, universities should redirect their energy toward adaptation. That means adjusting assignments, rethinking evaluation, and integrating AI use transparently so the focus remains on learning, not on detection.

Professors can start by building trust and treating students as partners rather than suspects. Many serious students still want to develop strong writing, communication, and critical thinking skills, especially the ability to read and write at an academic level. Instructors can tap into that motivation by designing tasks that AI can assist with but not complete on its own. They can ask students to compare chatbot drafts with their own revisions, explain how they used AI in their writing process, reflect on the strengths and weaknesses of AI-generated responses, or even participate in short oral exams. These steps can make AI use more productive and, most importantly, keep the focus where it belongs: on human learning. While it’s true that AI may have a greater impact than tools of the past, it can still be incorporated into the learning process if it is approached with care, creativity, and a clear purpose.

Artificial intelligence is here to stay, and it will only grow more powerful with time. Detecting it will become harder, its capabilities will expand, and its presence will become even more embedded in student life. But this shouldn’t be seen as a threat or the end of education as we know it. We saw how the rise of the internet brought about similar fears, yet it ultimately made learning richer and more accessible. Resistance to change is part of human nature, and large institutions like universities often move slowly. But whether they choose to lift AI restrictions or not, one thing is clear: The current “in-between” approach is failing both students and faculty. It’s time for schools to stop pretending this technology can be “defeated” and instead begin building an education system that works with AI, not against it.

This article AI is already in the classroom. It’s time colleges caught up. is featured on Big Think.

We’ve all experienced the illusion of knowledge. This cognitive bias leads us to walk around overconfident in the depth of our understanding of how the world works. We believe we know how zippers zip, the internet connects, and local politics work until someone asks us to explain. We then suddenly find ourselves fumbling for an answer that sounds plausible.

This happened to me last year when my son asked me to explain sanctions. I’ve been reading about them in the news for most of my adult life. Sanctions imposed on Iran to ice its nuclear program. Sanctions on Russia for invading Crimea. Sanctions on Venezuela for anti-democratic corruption. The list goes on, yet I barely stammered a respectable answer: “They’re economic penalties placed on a country to compel it to change something. I think they mostly have to do with the oil. So boats are probably involved … somehow. Maybe pipelines, too.” 

Okay, not exactly respectable, and truth be told, I left out the more embarrassing fumbles. Still, we don’t know what we don’t know until we know, you know? Thankfully, a good book provides a brilliant way to dispel the illusion of knowledge, so when I was given the chance to read Edward Fishman’s new book, Chokepoints: American Power in the Age of Economic Warfare, I took it.

Fishman’s book tells the story of how the United States turned the world economy into a battlefield by developing the economic weapons necessary to win international struggles before they expand into shooting wars. A leading authority in economic statecraft, Fishman works today as a professor at Columbia University’s School of International and Public Affairs; however, earlier in his career, he worked at the U.S. Treasury Department as a special assistant to the Under Secretary for Terrorism and Financial Intelligence at a time when the new rules for economic engagement were being written.

After reading his book, I sat down to talk with Fishman about this new age of economic warfare and what it means in our increasingly fractured, yet still globalized, world.

Economic warfare from Bosphorus to Baghdad

In defense of my stammerings, Fishman’s book opens with a boat. Many, in fact, all waging war across the 5th-century Aegean Sea. The skirmish pitted Athens against their Spartan rivals and marked the decisive moment in the 27-year-long Peloponnesian War.

With backing from the Persian Empire, which saw Athens as a mutual adversary, the Spartan leader Lysander constructed an armada large enough to destroy the ostensibly invincible Athenian navy. He then sailed victoriously to Lampsacus, a city strategically located on the Bosporus Strait. There, he seized the city and cut Athens off from the trade routes it relied on for its food supplies.

“The Bosphorus had been the Athenian’s lifeline, and the Bosphorus was where their empire met its demise,” Fishman writes.

The strait proved to be a chokepoint for the ancient Greek city-state — an economic gateway so critical that, as Fishman notes, “controlling it confers immense power” while “blocking it can bring your enemy to its knees.” Such chokepoints not only inspired Fishman’s title; they also served as the focal points of economic warfare for millennia. The Siege of Tripoli in the 12th century, Great Britain’s blockade of its rebellious colonies in the 18th century, the Union’s blockade of the rebellious South in the 19th century, and German U-boats prowling the Atlantic in the 20th century, all represent attempts to locate and squeeze a rival’s chokepoints in times of war.

And that’s how economic warfare went even up to the 1990s. To enforce international sanctions against Iraq for its invasion of Kuwait, U.N. coalition ships parked outside the country’s ports. For 13 lengthy and costly years, crews painstakingly inspected ships for banned goods. The most significant strategic update: Unlike in Lysander’s days, enforcement involved fines and impoundments rather than public executions in the city square.

As with so much else, everything changed after the 9/11 attacks. Iran’s efforts to develop a nuclear bomb stoked fears in Washington, yet the U.S. had already launched wars in Iraq and Afghanistan. The prolonged conflicts stretched the country’s willingness to enter another battlezone thin — and its credibility thinner after its justifications for Iraq proved unfounded. Meanwhile, Iran was getting closer to its goal.

“America was the most powerful country in the world, but here we were fighting these wars that were costing a lot of blood and treasure, but we didn’t have much to show for it,” Fishman said during our conversation. The prevailing question in some Washington circles became: “Is there another way for the U.S. to exercise its power and potentially get what it wants more effectively, cheaper, and less violently around the world?”

The Iron Bank of Davos

Enter Stuart Levey, a lawyer at the Justice Department who would go on to become the Treasury Department’s first undersecretary for terrorism and financial intelligence. Levey recognized that the 60 years since World War II had left the U.S. in a historically remarkable position. Its dollar had become the world’s reserve currency and the lifeblood of global trade. That and a lack of equally powerful geopolitical rivals gave the U.S. immense economic sway.

In hindsight, Levey’s plan seems obvious. The U.S. would go beyond sanctioning a few visas here and embargoing a specific export there. The country would also persuade the world’s banks to cease their business with Iran. CEOs would be shown declassified information about how Iran funded its nuclear program and terrorist proxies. They would then be given a choice: Cut ties with Iran or lose access to the all-important American dollar. The endgame was to isolate Iran from the international financial system and weaken the country’s economy. If enough economic pain accumulated, Iran would be compelled to negotiate without a single shot fired.

As Fishman writes: “Trying to navigate the global economy without access to the dollar is like trying to travel the world without a passport. […] The dollar’s importance in the global economy gave the United States a chokepoint of unparalleled strategy value.”

Of course, enlisting the private sector and other countries to support Levey’s plan would be no simple task. From this jumping-off point, Fishman’s book begins in earnest — though not in the way you might expect. His is no cold financial analysis nor a whittling down of the history to shape a clear moral lesson. Instead, Fishman chronicles the pivotal moments in this new era of economic warfare — from the Iranian sanctions to those imposed on Russia for its invasion of Crimea and the technological freeze-out of Huawei — by following the public servants who built America’s economic arsenal.

Yes, the odd graph makes an appearance (you get the sense talking with Fishman that he isn’t one to say no to a curvaceous area chart). But it’s the people behind the scenes who take center stage. The book even features a cast of characters that wouldn’t feel out of place in the appendix of a George R.R. Martin novel. It’s a welcomed inclusion given how many people come in and out of the various stories.

Readers explore this history through the eyes of people like Levey, David Cohen, and Daleep Singh. Thanks to Fishman’s thorough research and personal interviews, readers experience the dismay of an emerging complication, the relief of a long-delayed victory, or the frustration of a thwarted plan. These details create the atmosphere of a backroom political thriller without tipping into full-on drama. It remains a history book to its core.

Through this narrative choice, Fishman aims to demystify the confusion surrounding our new age of economic warfare for a wider audience. A more traditional analysis may give readers the knowledge to pass a multiple-choice test on these events. However, Fishman doesn’t want his readers to merely know the history of this new economic arsenal. He wants them to understand the people who built it and why they made the decisions that led history to play out as it did.

“It is very easy to go back and look at something that happened 10 years ago and say, “Oh, they were so stupid,’” Fishman says. “But the key thing is empathy. You want to understand what people knew at the time and the different competing pressures they were under. I thought that was the best way to tell the story in an accurate and engaging way.”

A poor statesman who blames his tools

Fishman is no polemicist, though. His goal isn’t to paint the U.S.’s economic arsenal as the savior of the free world nor as the smoking gun of an overreaching police state. He treats them as policy tools, pure and simple. Used wisely, they can prevent conflict from escalating into a war or humanitarian crisis. Used carelessly, the same tools can intensify international strife and weaken the case for bargaining.

To help me better understand the nuances, I asked Fishman what he considers an effective sanction versus an ineffective one. For him, the Iranian sanctions (2005–15) demonstrate economic warfare at its best. The goal wasn’t “economic pain for pain’s sake” but to target the precise pressure points that would successfully translate into diplomatic gains and eased tension.

“By the time Iran held a presidential election in 2013, their economy was reeling from these sanctions, and they effectively elected a president whose platform was, ‘We need to get the sanctions lifted.’ That led to [an interim] deal in late 2013 to freeze Iran’s nuclear program. This program, which had been inexorably advancing for almost a decade, is frozen without military force or a single loss of life in the United States.”

The Iran deal showed that sanctions could work and work effectively, ushering in what Fishman calls the “new age of economic warfare.” Over the last 25 years, each administration, Democrat and Republican, has roughly doubled the amount of sanctions as its predecessor. And if the first month of the second Trump administration is any clue, it may be more than double again by the time the next administration rolls in.

“You got a situation where officials at the Treasury or State Departments could sign a few documents and impose sanctions that were much more dramatic and impactful than when you used to park a ship outside someone’s port,” Fishman says. “It both supercharged the impact of sanctions while lowering the barriers to using them.”

Low barriers made applying unwise sanctions, including those imposed during the Venezuela crisis, easier. These sanctions began through congressional legislation during the late Obama administration and escalated during the early Trump years. The goal seemed to be repurposing the economic weapons used successfully against Iran and Russia to punish President Nicolás Maduro for his corrupt regime and violating human rights.

To be sure, Maduro is an authoritarian. He has been accused of stealing elections, committing crimes against humanity, and pursuing policies that have worsened the country’s already fraught economic troubles. However, sanctions have proven an ineffective tool for initiating regime change. As Fishman points out, leaders cut from such a cloth “are not going to commit political suicide to obtain sanctions relief for their people.” Instead, most of the blowback was felt by the very people Maduro oppresses, which only intensified the humanitarian crisis by further impoverishing the general population and arguably strengthening Maduro’s dictatorial grip.

“That’s when you run into real problems with sanctions: economic pain devoid of any realizable political objective,” Fishman adds. 

Big stick-and-carrot ideology

Fishman’s other hope for the book is to give readers the knowledge necessary to dispel their illusions and discuss these issues reasonably. That’s important because this new age of economic warfare won’t end soon. It’s just getting started, and as the last American election cycle made clear, things like sanctions, tariffs, and export controls will be at the forefront of foreign policy. 

“This is how great powers compete today,” Fishman says. “It’s not just the United States using these tools; it’s adversaries like China and Russia, and our allies like Europe and Japan. It’s paramount that we understand these tools because they are only going to become more important in the years ahead.”

Looking toward those years, I asked Fishman what might be in store as we leave the kumbaya geopolitics of the 1990s behind us. He envisions two potential scenarios. 

In the first, the United States strengthened its relationships with its closest allies: Canada, Mexico, Japan, and the European Union. Although the world economy has become far less globalized, integration remains vital. This future requires tending economic carrots to go along with our sanction sticks, and on that front, Fishman sees mixed signals.

The current Trump administration has proposed creating a sovereign wealth fund and supporting the country’s International Development Finance Corporation. However, the same administration has paused foreign aid and has gutted the Agency for International Development (USAID). That’s worrisome because not just economic weapons win wars before they begin. Disaster relief, development assistance, trade agreements, and environmental protections also help keep other states from tipping toward disaster while spreading diplomacy and American soft power.

This is the very reason President John F. Kennedy gave for creating USAID. As he said in 1962: “The people who are opposed to AID should realize that this is a very powerful source of strength for us. It permits us to exert influence for the maintenance of freedom. If we were not so heavily involved, our voice would not speak with such vigor. And as we do not want to send American troops to a great many areas where freedom may be under attack, we send you.”

It’s the second scenario that keeps Fishman up at night. Policymakers continue to “shoot from the hip” and use their economic arsenal to respond haphazardly and shortsightedly to crises without strategic planning. In this future, mistakes will pile up, tensions will escalate, and competition will spill out from the backrooms and spreadsheets.

“I worry about this not just for the economic reasons, and I think there would be significant inflation, unemployment, and economic pain,” Fishman says. “But I worry about it for the broader geopolitical ramifications because, as history has shown, if states can’t secure markets and resources through free trade, the temptation of conquest and imperialism will rise. I worry that today’s economic wars could become tomorrow’s shooting wars.”

This article How the United States turned the world economy into a battlefield is featured on Big Think.

A migrant today is anyone who crosses an international border. Beyond this basic definition, ways of classifying migration vary considerably. Does a day trip count? A month-long visit? Should students, tourists, and business travelers be considered migrants? To these questions can be added the usual problems associated with statistics, from poor record keeping to distortion by politicians.

What we do know is that the number of migrants worldwide has been rising steadily in recent decades, nearly doubling from 153 million in 1990 to 281 million in 2020, the most recent year for which the UN has published its global tally. As a share of the total population, however, there are not many more migrants today than in the past. The world’s population has increased by almost three billion in the past thirty years, meaning that the proportion of people migrating has remained relatively constant. In 2020, about 3.6% of all recorded citizens were born in a different country; thirty years earlier, it was 2.9%.

While this percentage could well fluctuate in the future, the overall number of people on the planet may be approaching its peak. The pace of global population growth is slowing after a period of rapid increase — from 2.5 billion people in 1950 to 5.3 billion in 1990, to today’s figure of 8 billion. The world’s population is expected to reach somewhere around 9.5 billion in the middle of this century, then fall to below current levels by the end of it.

A numbers game

Migration statistics should be regarded with great wariness. Not only are they notoriously difficult to collect and analyze, but they are also vulnerable to manipulation for political purposes.

Countries define and count migrants in different ways. Some measure a stay or period of absence from day one, others only after a year. Some add estimates of undocumented travelers to their overall tally of migrants, while others do not. There is no border control among the twenty-nine “Schengen” countries of Europe, for example, so a large amount of guesswork is necessarily involved. 

Even the richest countries are unable to measure migration accurately. The United Kingdom is one of many places that record entry and not departure. This results in significant discrepancies between reported figures for immigration and those for emigration. For example, Poland’s records of its migrants to the UK show far lower numbers than the British statistics reveal — and neither of the two is likely to be accurate.

Often figures vary wildly depending on who you ask. The Bulgarian Ministry of Foreign Affairs says that there were about 200,000 Bulgarians in the UK as of January 2019. Yet at the time the UK’s Office of National Statistics put the figure of Bulgarian-born people living in the country at 110,000.

Distinguishing between travel with the intent to settle and a journey made for tourism, business, or study is not easy, even with the help of specific visas, entry cards, and permits. We have much more accurate data on the flow of goods than on the movement of humans across borders.

Where are the migrants?

Of the estimated 281 million migrants in 2020, over two-thirds were workers, of whom over a half (59%) were men and two-thirds were working in high-income countries.

Europe is currently the main destination for migrants, hosting 87 million, or almost a third, of the total number worldwide. Very close behind is Asia, where there are an estimated 86 million migrants, or 30% of the total. About 59 million (20%) live in North America and 25 million (9%) in Africa — although the statistics regarding border crossings between Africa’s 52 countries are particularly unreliable, so this last figure is likely to be far higher.

In Latin America and the Caribbean, migration has doubled in the past fifteen years to 15 million people (5% of the global migrant population), although these too are­ significant underestimates given the scale of undocumented flows between Central American countries and recent emigration from ­Venezuela.

Geography and population sizes explain some statistical features. The extremely large, diverse, and sparsely inhabited region of Oceania has the highest proportion of migrants, with about 22% of those living in one country being born in another. By comparison, migrants constitute 16% of the population of North America and 12% of Europe’s.

The US is home to the largest absolute number of migrants — 51 million, or 15% of the total — with most originating in Mexico, India, China, and the Philippines, in that order. Over 11 million migrants have entered the US from Mexico, with the border between the two accounting for the most recorded crossings globally.

The US is followed by Germany, which has 16 million migrants, constituting 19% of its population. Saudi Arabia is the third-largest destination: the 13.5 million migrants living there make up 38% of its population, with India, Indonesia, and Pakistan as its top three origin countries.

As a share of its population, however, the United Arab Emirates hosts the most. About 88% of people living in the UAE are foreign-born. Around 3.8 million Indians work in the country, with over a million coming from Kerala and half a million from Tamil Nadu state.

Forced displacement

The UN Refugee Agency estimates that in the second half of 2023 there were 114 million migrants forcibly displaced due to persecution, conflict, violence, and violations of human rights. By the end of June 2024, that figure had increased to 122.6 million.

Of these, 35.3 million were refugees under the agency’s mandate, 5.9 million were Palestinian refugees registered with UNRWA, 62.5 million were internally displaced people, and 5.4 million were asylum seekers waiting for legal recognition of their need for protection.

Three quarters of all refugees were hosted by low-income and middle-income countries. The 45 least developed countries, as classified by the UN, were providing asylum to 20% of the total.

Most refugees flee to the closest place of relative safety. Nearly 70% of people in need of international protection live in neighboring countries. Turkey and Iran host the most refugees globally — about 3.4 million each, mostly from Syria in Turkey and from Afghanistan and Iraq in the case of Iran. The US receives the most individual applications for asylum, but it ends up hosting a relatively small number of refugees.

Together with the estimated 6.8 million displaced Palestinians, more than half of all refugees and others in need of international protection originate from just three countries — Syria (6.5 million), Afghanistan (6.1 million), and Ukraine (5.7 million). The conflict in Sudan is expected to add considerably to their number.

Some 1.7 million of these Palestinian refugees are living in Gaza. As a result of Israel’s military retaliation for the horrific events of October 7, 2023, their homes and communities have been destroyed. At the time of writing, over 30,000 had been killed, while many more had been injured and faced starvation as well as further violence, creating new waves of refugees.

In the three years since Russia’s invasion of Ukraine, more than 14 million Ukrainians have been forced to flee their homes, triggering one of the worst refugee crises since 1945. Their reception in much of Europe was in marked contrast to the responses to previous waves of refugees from other parts of the world. The fact that Ukrainians look much like other Europeans and are perceived to share closer religious and historical ties means they have enjoyed a warmer welcome than Syrian and Afghan refugees and been spared the hostility shown to African migrants.

The missing and the dead

Globally, the overwhelming majority of migrants make authorized border crossings. Undocumented or irregular migration, however, is by its nature difficult to quantify as it often involves clandestine, perilous journeys made by desperate people.

At least 60,000 people have died in the past ten years in the attempt to migrate to another country, with nearly half disappearing without trace, their bodies never found. In 2023, the deadliest year for migrants in the past decade, an estimated 8,565 people died trying to reach their destinations. The Mediterranean is perhaps the biggest graveyard for missing migrants, with tens and at times hundreds of thousands of desperate migrants attempting to cross every year.

A tiny fraction of the bodies of migrants who drown at sea are recovered; most of these “invisible shipwrecks” vanish without trace. Among those seeking to enter the US from Mexico, thousands disappear or die each year because of kidnappings, vehicle accidents, snake bites, dehydration, and starvation.

To leave or stay?

Around 97% of the world’s population do not become international migrants. Most people are not prepared or able to venture into the unknown. They may prefer the comfort of their familiar surroundings or be prevented from traveling abroad by poverty, age, ill health, or other restrictions.

The fact that there is vastly more migration within countries than migration between countries also reflects the difficulties of crossing borders and the desire of many people to remain close to home. It is hard enough to move within one’s own country; moving abroad is harder and riskier still.

Migration is often an enormous sacrifice taken on behalf of others. In many poor communities the eldest sons — or daughters, as is more often the case in the Philippines — may be encouraged to migrate to support their families. The perilous journeys, hardship, and loneliness migrants frequently endure on behalf of others is testimony to their self-sacrifice and courage. 

Refugees and other forcefully displaced people tend to travel as short a distance as possible so they can return when it is safe to do so. Anywhere from a fifth to a half of all migrants return home or move to a third country within five years. This may be because they have saved money, gained a qualification, or are coming back to settle, raise a family, or retire.

Migrants are unusually prepared to take risks and make sacrifices. These qualities prevented the extinction of our species during its early evolution, when threatened by droughts and famine. They lie at the heart of the extraordinary progress made by humans since. Though migrants are in the minority, we should not assume that being sedentary is somehow better or more natural. Migrants may be simultaneously very ordinary and highly exceptional people.

This article Numbers vs narratives: Unraveling the truth behind today’s global migration statistics is featured on Big Think.

This winter, while visiting my parents in California, I came across a YouTube ad that certainly wasn’t running in Utah, where I was living. Against stock footage of outdoor adventures and fresh produce, the narrator promised guilt-free living powered by “clean” electricity — all from hydropower. The message felt overly tidy, reassuring me that I didn’t need to think about where my electricity came from. Then the ad cut to drone footage of a massive dam — a visual reminder of how central dams are to hydropower.

That image jogged my memory. Just days earlier, I’d listened to a podcast about removing the Klamath River Dams in northern California. I’d gone camping and climbing in that area and remembered the dams and their impact on the landscape. The podcast celebrated their removal as a major win for the surrounding ecosystem, salmon runs, water quality, and Indigenous communities.

The contrast felt striking and obvious. On one side, dams symbolize clean energy. On the other, their removal represents ecological restoration. This raises the obvious question: Which is better for the environment — clean energy from hydropower or free-flowing rivers? And is it really an “either-or” narrative?

Conflicting desires or a shared mission?

When I described this perceived conflict to Malcolm Woolf, president of the National Hydropower Association (NHA), I heard a respectful but clearly exasperated sigh. “So much of public policy is driven by bumper stickers,” he said. “The notion that hydropower and dam removals are fundamentally at odds represents a false dichotomy.”

Katie Schmidt, associate director of the National Dam Removal Program at American Rivers, echoes Woolf’s sentiment.

“Most of the dams we’ve removed had no connection to energy production. In fact, fewer than 50 hydropower dams have been removed since the 1990s. When hydropower dams are removed, it’s usually because they’re no longer generating significant power or because maintaining them isn’t economically viable. Dams are only removed with the dam owner’s consent.”

Their messages were reassuring, but my confusion didn’t just stem from surface-level ads. It also came from the reality that hydropower hasn’t grown in decades. Meanwhile, dam removal, though gaining momentum, remains slow-moving. This imbalance could easily give the impression that the two efforts are fundamentally at odds — that progress in one comes at the expense of the other.

So, if the two aren’t opposing one another, what is slowing things down? Why aren’t we seeing more hydropower and more free-flowing rivers? 

It’s a shared struggle against a slow, complicated system that hinders progress on both fronts. Bureaucratic inefficiency plays a role, as does the inherent challenge of managing the world’s most precious resource: water.

“Water is inherently a finite, public good,” Woolf says. “As it flows downhill through our facilities, it’s used by other people in the community for different purposes. That means any decision involves a web of agencies, from federal to local.”

An “Uncommon Dialogue”

The surprising alliance between these two movements is rooted in a simple fact: The U.S. has built a lot of dams — around 90,000. Only 3% of them generate hydropower, while up to 85% are over 50 years old and in need of repair and management. However, experts like Schmidt say most don’t serve significant functions to justify the cost of repairs.

“The age of building major new dams ended 50 or 60 years ago,” Schmidt says. “Today, the question isn’t about building more — it’s about what to do with the 90,000 we already have. Removing obsolete dams doesn’t reduce energy production. Instead, it opens the door for healthier rivers, improved ecosystems, safer communities, and cleaner water — all while leaving plenty of existing dams available for hydropower expansion.”

Existing dams can also be retrofitted, a process that can be less ecologically harmful than constructing entirely new energy projects like wind or solar farms. That’s because the candidate dams are already part of the landscape and serve critical functions such as flood control, irrigation, or shipping. Their environmental impacts, such as disrupting natural river flows or fragmenting habitats, exist regardless of whether they generate energy. By retrofitting these structures to produce hydropower, we add renewable energy capabilities to infrastructure that would otherwise continue to exist without providing additional benefits.

In other words, retrofitting turns an ecological liability into a dual-purpose asset: The dam continues its original function while contributing to the nation’s renewable energy goals.

This idea led to the “Uncommon Dialogue,” a collaboration between hydropower advocates and dam removal proponents. Facilitated by Stanford University’s Steyer-Taylor Center for Energy Policy and Finance and the Stanford Woods Institute for the Environment, the dialogue brought together seemingly opposing groups to find common ground. Their focus? Remove ecologically harmful dams while retrofitting others for clean energy. To do so, they have to tackle shared challenges, namely the complex regulations governing water.

The state of dam removals: Slow flowing

Since 1999, more than 1,900 dams have been removed in the U.S., restoring thousands of miles of rivers. These projects have revived fish habitats, improved water quality, and boosted recreational opportunities. Often, they demonstrate that dam removal benefits the environment and local communities alike.

However, scientists have studied the impacts of fewer than 10% of dam removals, with even less research on long-term effects. Outcomes can vary depending on factors such as dam type, removal methods, surrounding land use, and watershed conditions. While most experts agree that dam removals often provide environmental and community benefits, the specific impacts differ case by case, and more research is needed to fully understand their long-term effects. 

On average, the United States removes 60 to 80 dams a year, a number Schmidt hopes nudges closer to 100 over the coming years. But there are significant roadblocks that slow down progress.

First, dam removals are complex undertakings that involve navigating state and federal permitting, lengthy fundraising efforts, and detailed hydrology assessments to evaluate post-removal water flow and environmental impacts. If a dam is federally regulated for hydropower, the dam owner must formally surrender their license. Schmidt notes that this process alone can take two to five years, is costly, and doesn’t always result in removal.

Beyond regulatory obstacles, misinformation creates resistance to dam removal. Some community members fear removing a dam will cause floods or leave permanent mudflats. Schmidt explains that only 17–20% of dams provide flood control because they require large reservoirs. As for mudflats, they do emerge briefly after removal, but proper restoration allows vegetation to return quickly. “We’ve seen dormant seeds sprout almost immediately after water levels drop,” Schmidt adds.

Another pervasive myth is that dams make more water. Schmidt clarifies, “Dams manage water, but adding one doesn’t mean you inherently have more water — it’s about controlling the flow and storage of what’s already there.” This misconception, especially prevalent in water-scarce states like California, often leads to misguided policies.

The state of hydropower: Stalled out

If dams suitable for hydropower retrofits exist, why isn’t hydropower growing?

“Hydropower retrofits don’t fail because of money or technology,” Woolf explains. “Companies are ready to invest in the infrastructure, sell the power, and recoup their costs. What slows us down is the regulatory process and a relicensing process that makes surrendering a license sometimes a more feasible alternative for owners.”

In 2018, Oak Ridge National Laboratory identified 100 existing dams with the highest potential for hydropower retrofits. Of those, 88 are owned by the U.S. Army Corps of Engineers, whose priorities often lie in flood control, navigation, and water supply — not energy production. That said, the Corps isn’t opposed to hydropower retrofits. In fact, they are the largest single contributor to clean renewable power in the country, simply because they manage so many hydropower facilities. 

“Retrofitting existing dams for hydropower requires threading the needle between competing priorities,” Woolf says.

Then there’s the licensing process. Every 50 years, hydropower plants must renew their licenses with the Federal Energy Regulatory Commission (FERC), a process that can drag on for eight to ten years. The primary cause? The sheer number of agencies involved in examining key questions one after the other. Woolf compares the process with the nuclear industry: “They have one regulatory body that aims to complete reactor relicensing in 18 months.”

With 30% of hydropower dams up for license renewal in the next decade, some facility owners might opt to surrender their licenses rather than endure the costly, decade-long process. This could trigger a wave of dam retirements, throwing us backward, not forward, in hydropower.

Some argue that dam removals and hydropower relicensing should take as long as they do, given their potential for a huge impact. Woolf says we can speed up the process while still being thorough.

“In particular, every agency with a stake in the dam removal or retrofit process waits for the others to finish their reports before they start their own,” he explains. “We’re essentially doing things in series, not together. Take every issue related to a dam — fish, water quality, recreation — and every agency runs their report one at a time. First FERC, then NEPA [the National Environmental Policy Act], then the state Fish and Wildlife Department. It’s all important, but if these agencies worked in concert instead of sequentially, we’d save years off the process without sacrificing thoroughness.”

And the complexity of balancing these priorities and navigating federal processes slows progress. 

“Since that 2018 report, only two or three of the top 100 dams have been retrofitted,” Woolf says. Meanwhile, alternative energy options like wind and solar can be developed much faster. “In the time it takes to retrofit a dam, you could build and take down a wind farm.”

The costs of leaving outdated dams

While the U.S. drags its feet on hydropower, its electricity demand is skyrocketing — fueled by data centers, air-conditioning needs, and population growth. Hydropower, as the only renewable energy source that can provide 24/7 power, is uniquely positioned to meet this demand. Unlike solar and wind, hydropower can be adjusted to meet demand at any time; it doesn’t rely on the wind blowing or the Sun shining. 

Without retrofits, natural gas is stepping in to fill the gap, locking in decades of fossil fuel dependence.

In 2016, the U.S. Department of Energy released a report with a name as optimistic as those hydropower YouTube ads: “The Vision of Hydropower.” In it, they described a future where we get more and more renewable energy from hydropower and pumped storage facilities. The report estimated that hydropower capacity could grow from 101 GW to 150 GW by 2050 — a nearly 50% increase. The latest statistics, nearly a decade later? In 2023, hydropower was around 80 GW. Clearly, we are not meeting that goal.

Leaving outdated dams in place comes with its own costs: degraded ecosystems, threats to public safety, and missed opportunities for river restoration.

Where there’s a sill, there’s a way …

The potential of hydropower and dam removal working together has concrete success stories, like the  Pawcatuck River project in Rhode Island. There, outdated dams were removed to restore free-flowing river ecosystems while increasing hydropower generation at other sites. By taking a basin-wide approach, they restored over 100 miles of wild river and managed to enhance renewable energy production at the same time.

There are also efforts to streamline the licensing process. In 2023, hydropower and dam removal advocates, alongside other stakeholders, introduced the Community and Hydropower Improvement Act (CHIA) in Congress. The bill aims to accelerate federal permitting for dam removals and hydropower retrofits by reducing redundant agency reviews while preserving environmental protections. With its focus on cutting red tape, CHIA could appeal to the new Republican-led House and Senate, but it has stalled amid broader legislative gridlock and competing priorities.

It’s an issue of vision and political will. “We don’t build these projects for one four-year term; we build them for decades. With the proper maintenance, these are forever assets” Woolf says. “It’s not about choosing one over the other; it’s a rare opportunity where doing both will benefit the environment and help decarbonize the grid.”

As for the future painted by the hydropower ad, well, I still don’t think most of my electricity will come from renewables anytime soon. But it’s reassuring to know we don’t have to choose between healthy rivers and hydropower — in other words, we’re not dammed if we do and dammed if we don’t.

This article The surprising alliance shaping the fate of America’s 90,000 dams is featured on Big Think.

Something is happening in the lunchrooms of America’s middle and high schools: They’re getting louder.

In the past few years, the aural environment has been dominated by clanging trays, pinging registers, and scraping silverware. Now, raised voices are once again soaring above the cacophony. Educators are generally happy with the trend. It’s an audible sign that smartphone bans are working. No longer glued to their screens, students are speaking to each other again!

Adam M. McCready, an assistant professor-in-residence of higher education and student affairs at the University of Connecticut, hears this anecdote over and over. He and his colleagues are exploring the effects of smartphone bans in public schools, but a handful of anecdotes does not constitute convincing scientific evidence. As policymakers barrel ahead with purging smartphones from classrooms, will these efforts benefit students?

The social media generation

For the past decade and a half, children have been guinea pigs in an unprecedented social experiment. What happens when they are given powerful, pocket-sized personal computing and communication devices outfitted with addicting software with relatively little supervision? Can their immature prefrontal cortexes handle such distractions, especially in educational settings? Now that 95% of American teenagers have access to a smartphone and spend hours on it daily, the consensus among adults appears to be a resounding “no.”

Overall, 68% of US adults responding to a 2024 Pew Research Center survey say they support a ban on middle and high school students using their phones during class. The biggest reasons these Americans gave for backing the move? Fewer distractions (98%), better social skills (91%), less cheating (85%), and reduced bullying (70%).

Buffeted by resolute public opinion, politicians are taking action. As of November 4, 2024, eight states — California, Florida, Indiana, Louisiana, Minnesota, Ohio, South Carolina, and Virginia — have passed policies that ban or restrict phone use in schools. These broadly take effect during the current school year. Twelve more states have introduced such legislation.

The movement to ban smartphones in schools extends globally as well. France, Spain, Switzerland, Canada, Israel, and Sweden have all introduced policies at a national level, with Chile, Denmark, and England on the same trajectory. In 2023, the United Nations Educational, Scientific, and Cultural Organization (UNESCO) called for a global smartphone ban in schools to “tackle disruption by smartphones in classrooms and protect children from cyberbullying.”

But just as putting smartphones in teens’ pockets was a grand experiment, so too is taking the devices away. Parents, educators, and policymakers hypothesize that restricting phone use will improve teens’ social lives, boost academics, and cut down on bullying. Does science bear this out?

Unconvincing early research

The evidence collected thus far is unconvincing. Any positive effects of restricting smartphone use in middle and high schools are small or nonexistent.

Schools started to implement smartphone bans in the wake of the COVID-19 pandemic. Researchers have taken these opportunities to compare metrics before and after policies are enacted or, more commonly, to weigh outcomes in schools that ban smartphones versus schools that don’t.   

In one review published in August, researchers at the University of Augsburg in Germany aggregated and analyzed five research studies looking at schools in Europe. Combined, the research suggests that bans can reduce social problems, the reviewers found. With reduced access to social media during school, kids couldn’t cyberbully each other as much. The reviewers also discovered a slight boost to academics.

Based on the early research, they recommended that “smartphone bans be introduced in schools, accompanied by educational measures, and evaluated regularly.”

In a widely touted study published in February 2024, Sara Abrahamsson, a postdoctoral researcher with the Norwegian Institute of Public Health, looked at the effects of smartphone bans in schools in Norway. She found that bans greatly boosted girls’ mental health and slightly augmented their grades, while boys were unaffected. Abrahamsson theorized that girls’ higher phone use compared to boys could account for the difference in effect. 

In another study carried out in South Australia, researchers compared schools that had instituted a ban requiring students to keep phones off and away during the entire school day versus those without such a policy.

“Overall, the results indicated that the ban and no ban schools either did not differ significantly, or there were minimal differences, in terms of problematic use of mobile phones, academic engagement, school belonging, and bullying,” the authors reported.

Marilyn Campbell, a professor in the School of Early Childhood and Inclusive Education at Queensland University of Technology in Australia, has conducted the most scoping review of the research on smartphone bans to date. She and her coauthors looked at 22 published and unpublished studies carried out across the world. Their assessment wasn’t exactly glowing.

“Overall, our study suggests the evidence for banning mobile phones in schools is weak and inconclusive,” they concluded. Any potential improvements to academics, mental health and wellbeing, and cyberbullying don’t seem to be meaningful. 

They added that while one-size-fits-all policies at state or national levels may be politically popular, they aren’t in the best interest of schools or students. “Politicians should leave this decision to individual schools, which have direct experience of the pros or cons of a ban in their particular community.”

In general, the studies on smartphone bans were also poorly designed and could not root out cause and effect, Campbell and her coauthors added. 

Moreover, policies have proven to be uneven, with bans taking various forms. Some schools have students store phones inside backpacks for the day. Others use safe-like portable storage units or have students lock phones away in magnetically-sealed pouches. Some simply install classroom phone holders on desks where all students place their devices.

Although the benefits of smartphone bans aren’t apparent for students, teachers’ desires must be considered as well. Nearly three-quarters of high school teachers, a third of middle school teachers, and even 6% of elementary school teachers say that “students being distracted by their cellphones in the classroom is a major problem,” according to a Pew Research report published in April.

Here, the Digital Wellness Lab at Boston Children’s Hospital declares the science to be clearer: ”Most classroom cell phone [sic] bans are seeking to reduce the distractions presented by unstructured use of cellphones, such as social media pings, text messaging, and video games. On this topic, evidence for the negative effects of media multitasking is more clear than some of the other facets of a smartphone ban.”

Don’t ban, educate

McCready broadly agrees with Campbell and her co-authors. As he told Big Think:

“The efficacy of any smartphone ban is context dependent. We have observed that some schools that claim publicly to have implemented full-day bans are not implementing these bans with the intended fidelity (e.g., phones are not secured at the start of school). Just because a ban is in place does not necessarily mean that students are unable to access their devices or that they are not misusing school provided technology.”

McCready and Campbell also concur that educating teens about proper smartphone use is likely a better, longer-term solution compared to blanket bans. “This will help students learn how to use their phones safely and responsibly at school, at home and beyond,” Campbell wrote.

“My research team believes that fostering digital citizenship, literacy, and agency among students is vital to student well-being and success,” McCready said. “The best analogy I can give is related to driving a car. In the United States, we require teens and adults to take driver’s education courses, possess learner’s permits, and take driving tests before they earn their licenses. However, many children receive smartphones, iPads, Chromebooks, and other devices with little or no education or guidance.”

A ban simply reinforces the notion that phones are universally harmful, he says, which isn’t true. They’re a tool, one which requires instructions on proper use.

“We observed students in schools with bans immediately access their devices at the end of the school day. Their online behaviors outside of schools go unchanged,” McCready said.

Adults would probably benefit from lessons on smartphone hygiene as well, he added.

“The public discourse also often fails to acknowledge how adults model the use of smartphones and other devices for children. How are educators using their devices during the school day in front of children? How are parents, caregivers, older siblings, and others modeling these behaviors at home? If children’s smartphone use is a concern, then we need to consider adult smartphone use as well.”

A new moral panic?

Take a step back and zoom out. In many ways, the rising anti-phone fervor in schools mirrors moral panics of the past. Politicians and parents have invariably blamed all sorts of things — rock ‘n’ roll, Marilyn Manson, Dungeons & Dragons, violent video games, and so on — for corrupting the youth. Social media-equipped smartphones could simply constitute the next target in elders’ crosshairs. Again, scientific evidence suggests these new technologies aren’t inherently good or bad, and their effects on teens are nebulous.

But of course, the topic of smartphone bans in schools is far from settled and clearly merits further, higher-quality research. Professor McCready aims to provide. 

“To my knowledge, researchers have yet to parse out which methods are most effective,” he told Big Think. “My mixed methods study with Drs. Kathy Rohn [UConn] and Jennifer Phaiah [Sacred Heart] is investigating the efficacy of the using Yondr pouches to secure phones for the entire school day. We are collecting longitudinal data from seventh-grade students from four middle schools at three timepoints during the current school year. Two schools have implemented Yondr pouch bans, while the others have not. In addition, two schools (one with pouches, one without) have implemented the Common Sense digital citizenship curriculum.”

The results of the study should be available next fall.

This article US barrels ahead with school smartphone bans — despite weak evidence that they help is featured on Big Think.

This is T-Minus, where Freethink counts down the biggest developments in space. For the last installment of 2024, Kristin Houser looked back on the 10 greatest space stories of that year. From a first-of-its-kind spacewalk to the recovery of a rocket that could one day take people to Mars, here’s what she found.

SLIM’s moon landing

In 1966, the USSR became the first nation to achieve a “soft” lunar landing, meaning one in which the spacecraft isn’t destroyed upon impact. On January 19, 2024, Japan became just the fifth nation in history to do the same, soft landing its Smart Lander for Investigating Moon (SLIM) spacecraft on the lunar surface. The landing wasn’t perfect — SLIM hit the moon upside down — but the location was the most precise in history, with the spacecraft touching down within 180 feet of its target.

Ingenuity’s final flight

In 2021, NASA’s Ingenuity helicopter made history as the first craft to fly on another planet. The original plan was for it to fly no more than five times over the next 30 days, but by the time the overachieving little chopper’s mission finally ended on January 18, 2024, it had flown 72 times, repeatedly setting new height, speed, and distance records and foreshadowing a future in which our space landers and rovers are joined by eyes in the sky.

Return of the space factory

California-based startup Varda Space Industries is planning to manufacture products in orbit, mainly specialized materials that could benefit from being made in free fall, without the pressure of Earth’s gravity. In June 2023, it launched a demonstration mission with the first of these “space factories,” and after a lengthy delay, the spacecraft finally returned to Earth on February 21, 2024, with a successfully synthesized drug on board — potentially signaling the start of a new era of off-world manufacturing.

The Polaris Dawn spacewalk

In 2021, billionaire entrepreneur and experienced pilot Jared Isaacman commanded Inspiration4, the first all-civilian space mission. In September 2024, he returned to Earth’s orbit with another civilian crew in the Polaris Dawn mission, which saw all four crew members exit their SpaceX Dragon capsule to perform the first private spacewalk. Next for Isaacman? Potentially serving as the new head of NASA.

Vulcan Centaur’s first flight

In 2014, United Launch Alliance (ULA) — a joint venture between Boeing and Lockheed Martin — began development on Vulcan Centaur, a new heavy-lift rocket designed to succeed its workhorse Atlas V launch system. After multiple delays, the rocket finally lifted off for the first time on January 8, 2024. A second successful flight followed in October, and the rocket is now certified to begin flying national security missions for the US Space Force in 2025.

Intuitive Machines’ moon landing

Vulcan Centaur’s primary payload for its maiden launch was Astrobotic Technologies’ Peregrine lunar lander, which was expected to be the first spacecraft built by a private company to reach the moon. Unfortunately, its propulsion system failed soon after deployment. But less than a month later, another private company, Intuitive Machines, deployed its own moon lander, Odysseus, via a SpaceX rocket, and after a harrowing descent, this one did reach the lunar surface — earning a place in commercial space history.

The launch of Europa Clipper

Scientists suspect that Jupiter’s moon Europa has liquid water, organic compounds, and a source of energy — the three ingredients needed for life as we know it. That makes the little moon one of the most intriguing places in our solar system, and on October 14, 2024, NASA launched Europa Clipper, a mission to study the icy moon up close. In 2030, after travelling 1.8 billion miles to reach its destination, the spacecraft will perform 49 flybys of Europa to help us determine whether life could exist beyond our home planet.

Chang’e 6’s round-trip moon mission

In 2019, China became the first nation to soft land on the far side of the moon (the side permanently facing away from the Earth). It repeated the feat with its Chang’e 6 lander on June 1, 2024, but this time, it equipped the spacecraft with a module that could launch samples of far-side lunar regolith to a lunar orbiter. A return vehicle attached to the orbiter then brought the samples to Earth — giving us our first look at the composition of the “dark side” of the moon. 

Starliner’s troubled ISS flight

In 2014, NASA issued billion-dollar contracts to Boeing and SpaceX to build capsules that could carry astronauts to and from the ISS. SpaceX delivered with Dragon in 2020, but it wasn’t until June 5, 2024, that Boeing was finally ready for the first crewed flight of its Starliner capsule. 

The vehicle did reach the ISS the following day, but experienced issues with its thruster and helium systems along the way. After extensive delays and attempts at troubleshooting, NASA ultimately decided it was too risky to have astronauts Butch Wilmore and Suni Williams use the Starliner capsule for their return flight to Earth, so instead of spending a week on the ISS, they’ll be staying there until February 2025 — and then hitching a ride home with SpaceX.

SpaceX’s stunning “Mechazilla” recovery

After more than a decade of development, SpaceX finally flew its massive Starship rocket for the first (and second) time in 2023. Both flights ended with the rocket exploding because of issues that triggered its autonomous flight termination system — but it did reach space on its second flight, earning the top spot on last year’s list of the biggest stories in space.

This year, SpaceX flew Starship four more times, and during each flight, it was able to test new systems and capabilities. The most remarkable of those tests took place on October 13, when SpaceX flew Starship for the fifth time and then used mechanical arms on its giant launch tower, dubbed “Mechazilla,” to catch its Falcon Heavy booster after it returned to Earth.

Video of this first-of-its-kind rocket recovery is a must-see for anyone with a remote interest in aerospace, and its success put SpaceX another step closer to not only using Starship for crewed missions to Mars, but doing so at a cost far lower than would be possible with an expendable rocket system.

This article T-Minus: Counting down the top 10 space stories of 2024 is featured on Big Think.

It’s a summer day in August 2017. Behind the quiet sand dunes near Petten, a village in the Netherlands where the North Sea is kept at bay, the only noise comes from a group of seagulls. They’re fighting over the remains of a lunch someone has left behind. Other than that, the place is deserted. You’d be surprised to find that some hold this as a historic place and a historic day.

Here in Petten, at the facility of the NRG (Nuclear Research & Consultancy Group), a project has just begun what could be a milestone in the development of nuclear power. As of this morning, the research reactor contains six lumps of salt, each the size of a die and carrying a few grains of thorium. It may not sound that spectacular. Yet thorium has been heralded as an improvement in nuclear technology, and could be the next step in the transition to a clean energy supply. Thorium is a weakly radioactive element, which is even more abundant in the Earth’s crust than uranium. A so-called molten salt reactor using thorium produces waste that’s relatively short-lived, ‘only’ 300 years or so. The reactor is supposed to be entirely safe thanks to features such as a plug of solidified salt, which melts automatically when the reactor gets too hot, for instance if the cooling pumps lose power. Once the sealing plug has melted, the salt flows into storage vessels underneath the reactor and the nuclear fission process stops. 

Put simply: since the fuel is already molten, nobody needs to worry about a meltdown. If the reactor overheats, things will be fine, not thanks to human intervention but to the laws of nature.

Over a cup of coffee in the canteen, Sander de Groot, a product developer at NRG, can’t stop talking about thorium. The first time he read about it must have been around 2006, he says. A molten salt reactor containing thorium had been running at the Oak Ridge National Laboratory in the 1960s. It worked. Everything that went on with the reactor was meticulously recorded at the time. For decades, these reports had been gathering dust in a filing cabinet. At the beginning of this century, a young employee of NASA rediscovered the reports, scanned them and made them available online. De Groot stumbled upon them a few years later.

Whenever De Groot talked about thorium with colleagues, they found it “too wild” and “speculative,” he says. A smile appears on his face. “Well, today, at last it’s happening.” A lot has changed since — and not just for Sander de Groot, who moved on in 2022 to co-found Thorizon, a start-up designed to accelerate developments in the use of thorium. First there were the number-crunchers, checking to see if such a reactor could operate without any problems. Then came the entrepreneurs, looking for investors to support their business cases. And, now, it’s time for the next step: fundamental research, financed with public money and carried out in a reactor owned by the European Union. The Netherlands is leading the way. Delft University of Technology is coordinating an international collaboration to develop the design of a thorium molten salt reactor.

“The potential is huge,” says De Groot. “In a molten salt reactor, thorium can be used so efficiently that it can be a safe and almost infinite source of energy.”

After decades of stagnation, a wave of innovation is sweeping through the nuclear industry. Thorium is one of many options for the future. Work is being done on fuels that can withstand every conceivable accident, and on reactors that can more easily cope with the erratic, weather-dependent power production of wind turbines and solar panels. Much of the focus is on small modular reactors (SMRs), for which components are produced in factories and assembled on site. Industry representatives say new nuclear plants are becoming more efficient, requiring less uranium and producing less waste. Or they are becoming safer, thanks to yet another layer of measures. Some designs are intended for sites near coastal cities, others for remote areas. Some use both the power and the heat released to produce hydrogen, which can be used in industry or as fuel for aircraft and container ships.

And then there’s nuclear fusion. In a fusion reactor, temperature and pressure are extremely high so atoms squeeze together like they do in stars, releasing an awful lot of energy. So far, nuclear fusion has always been something that exists only on the horizon. In 2021, a milestone was reached at the Oxfordshire-based Culham Centre for Fusion Energy where researchers managed to sustain fusion for five seconds, releasing 59 megajoules of energy, enough to boil a few dozen kettles of water. It’s also enough to get dozens of headlines published across the world.

All these stories feel like a breath of fresh air. Could the nuclear industry finally be changing? Is there room for new and bold ideas that give nuclear energy a future? Might it be, after its dark journey through Hiroshima and Chernobyl, on its way to an awakening?

In the emerging pro-nuclear movement, there’s plenty of discussion about the benefits and downsides of each new reactor design. There’s a catch to everything. In one design, the material would be prone to corrosion, while in another, disposal of fission products may become a concern. Breeder reactors could dispose of nuclear waste, but they might also be used for the production of weapons material.

With their technical disputes over the pros and cons of each reactor, proponents of nuclear power inadvertently reinforce the main point of their opponents: that today’s nuclear plants are flawed. ‘Advanced nuclear’, the generic term for all innovations, suggests that current nuclear plants are not advanced. It’s as if they’re obsolete and in need of serious adaptations.

And why is there so much focus on innovation, really? Nuclear plants have a pretty good track record. The problems with nuclear power have nothing to do with technology. The problems are between the ears.

Nuclear plants are distrusted, so now, with each innovation, they need to be made socially acceptable. Safer, because nuclear power is fraught with dangers. Flexible, because solar and wind should be prioritized. Smaller, because small is beautiful. But will all this innovation really make nuclear power better? Does anyone believe that a different type of reactor or fuel will make resistance disappear?

This article Why the golden unlock with nuclear power will be mental is featured on Big Think.

It’s 2030, and more than two-thirds of the population of sub-Saharan Africa is now online. While this internet access is helping residents combat poverty, disease, illiteracy, and more, continuous effort will be required to ensure the benefits of the tech outweigh its potential harms.

Africa’s “digital decade”

It’s been 55 years since a team of researchers at UCLA sent a message from their university’s room-sized computer to another massive machine housed at Stanford, giving birth to the technology that would eventually become the internet.

Since then, computers have gotten small enough to carry around in our pockets, the wires are optional, and two-thirds of the world’s population is now online. 

For individuals, this internet access has meant opportunities in work, education, healthcare, and more that aren’t available to those still offline. On a national level, digital connectivity has typically translated to a more productive workforce, higher standard of living, and economic growth.

These benefits are currently out of reach for the majority of people in the nations of sub-Saharan Africa, where the internet adoption rate is currently just 37%, but tech giant Google is predicting that the next 10 years could be the region’s “digital decade.”

To find out what that could mean for sub-Saharan Africa, let’s look at the history of digital tech on the continent, how Google is working to bring its own prediction to fruition, and what can be done to help ensure that internet access is a net positive for Africa’s most vulnerable residents. 

Where we’ve been

Where we are

Sub-Saharan Africa — the part of the continent below the Sahara Desert — doesn’t just have the world’s lowest internet adoption rate. It’s also home to the majority of the world’s least-developed countries (LDCs).

These nations are characterized by low GDPs and populations experiencing high levels of poverty, malnutrition, illiteracy, gender inequality, and other barriers to sustainable development. Internet access could help break down these barriers, but LDCs have particularly low internet adoption rates, with just 27% of their residents online.  

Internet access isn’t the only area where LDCs are lagging behind. Progress in technologies enabled by the internet, like AI, is advancing rapidly in developed nations, and according to the UN, the gap between the countries with the most advanced tech and those with the least is actually widening.

Getting sub-Saharan Africans online is the first step to closing that gap, and despite its still-low internet adoption numbers, the region is trending in the right direction — a Google-funded report by Public First, a public policy research agency, has predicted that more than half the area’s population will be online by 2026.

“The next decade is set to be sub-Saharan Africa’s digital decade — with emerging technologies set to significantly accelerate the continent’s development,” Matt Brittin, president of Google Europe, Middle East, and Africa, predicted in an October 2024 blog post.

Where we’re going (maybe)

To close the digital divide, the region first needs affordable, high-speed internet access. Its residents then need to be taught the skills needed to take advantage of the digital world.

Since opening its first office in sub-Saharan Africa in 2007, Google has invested millions into helping bring both access and education to the region — it’s partnered with manufacturers to make cheaper smartphones for the African market, helped millions of Africans access digital tools and training, and built Equiano, a subsea cable speeding up the connection between Africa and Europe. 

In 2021, the tech giant announced its biggest initiative yet: a $1 billion investment over the next five years into Africa’s “digital transformation.”

“This research reaffirms the importance of our investments.” 

Alex Okosi

Google says it has already spent $900 million on projects in Africa in the three years since announcing the investment, and according to the report it commissioned from Public First to look into its impact on sub-Sharan Africa, it’s on track to meet the $1 billion goal by 2026. 

Among other things, the investment money has been used to open new AI labs in Ghana and Kenya, fund initiatives designed to support African founders, and create Africa’s first “Google Cloud” region and its first Google product development center.

“This research reaffirms the importance of our investments,” Alex Okosi, Google’s managing director for sub-Saharan Africa, wrote in the report’s forward. “Public First found that every $1 invested in digital technology in the region will generate over $2 in economic value by 2030, with an even greater return in front runner nations.”

In May 2024, Google announced its next big project as part of the investment: building Umoja, the first subsea fiber-optic cable between Africa and Australia. According to Okosi, this will help address the two biggest barriers to closing the digital divide in the region: high data costs and connectivity challenges.

“Investing in infrastructure like Equiano and Umoja are going to help bring down these costs and enable businesses to hopefully tap into this digital highway that enables them to be more efficient,” he told CNN.

“We still need to partner with last-mile operators to make sure that they are leveraging this new bandwidth, and it’s bringing the cost down so the consumer can feel it,” Okosi added. “We don’t want to leave anyone behind, and the only way to ensure that is to make sure the cost of data is lower.”

In addition to sharing the update on its $1 billion investment, Google also recently announced that its philanthropic arm, Google.org, is committing $5.8 million in grants “to support AI skilling and education across sub-Saharan Africa.” 

This money will help ensure that Africans are able to take advantage of the technology once they have access to it, and the Raspberry Pi Foundation, a UK-based charity focused on the promotion of computer science education, is one of its recipients.

“In Africa, we are partnering with Young Scientists Kenya and Data Scientists Network Foundation to roll out AI literacy education for Kenyan and Nigerian youth,” a spokesperson for the Foundation told Freethink of its plans for the grant. 

“Young people need to be more than just consumers of technology. They need to be digital protagonists, able to confidently engage with and shape the digital world.” 

A Raspberry Pi Foundation spokesperson

This education will include lessons on different types of AI, how machine learning can be used to solve problems, and the technology’s limitations. These lessons will be tailored to “students aged 11-14 with a focus on those from underserved communities,” according to the Foundation spokesperson.

The Raspberry Pi Foundation is also launching several new resources focused on AI safety in Africa. One of these, “AI and Your Data,” is designed to get students thinking about how they share data with applications in the age of AI, while another focuses on how AI can be used to both perpetuate and combat misinformation.

“Young people need to be more than just consumers of technology,” the spokesperson told Freethink. “They need to be digital protagonists, able to confidently engage with and shape the digital world.” 

Avoiding the pitfalls

This idea of Africans as tech consumers highlights the primary incentive Google and other tech companies have for helping close the digital divide in places like sub-Saharan Africa: more people online means more people using their products, which ultimately means greater profits.

“The big US tech giants have recognized the existing connectivity gaps and the need for additional investment associated with this as a major business opportunity,” Tevin Tafese, a data scientist at the German Institute of Global and Area Studies, told German broadcaster DW.

“Prominent examples are Google and Meta, whose major cable projects are aimed at reducing the cost of accessing their own service in a largely untapped African market,” he continued.

This investment might seem like a win-win on the surface — Africans gain access to the digital world, while tech companies gain access to more customers — but the reality may be more complex, given the downside of the internet and modern tech.

“Virtually every day brings new stories about hatred being spread on social media, invasion of privacy by businesses and governments, cyber-attacks using weaponised digital technologies, or states violating the rights of political opponents,” the UN’s High-level Panel on Digital Cooperation wrote in a 2019 report. 

These impacts can be felt by tech users across the globe, and it’s not clear how we can effectively address them in developed nations like the US, let alone in regions like sub-Saharan African, where resources are scarcer and much of the population is already incredibly vulnerable.

Joining the digital world also puts people living in LDCs at risk of experiencing “digital extractivism,” a practice that includes the exploitation of data and labor, according to a Stanford-funded report by researchers at Pollicy, a Uganda-based civic technology organization.

“Technology companies have exploited minimal data protection legislation and local competition to set up widespread software communication solutions for the African market to collect user data,” the authors write.

They also “hire digital workers from locations, such as within Africa, where hiring labour is cheaper or where labour laws might be weaker in a practice referred to as labour arbitrage,” Pollicy’s researchers add.

There are ways to prevent this exploitation and protect the Africans now coming online, according to Pollicy: campaigns can be organized to raise consumer rights awareness among new internet users, for example, and the strengthening of trade unions can help grant digital workers more bargaining power. 

Ultimately, though, while the US and other developed nations might have a headstart on places like sub-Saharan Africa, the internet, AI, and other digital technologies are still relatively new inventions, and we’re all trying to figure out how to maximize their benefits while minimizing their harms — a particularly daunting task given how fast the tech is evolving.

As we navigate the ever-changing landscape of this digital world, stakeholders will need to take special care to protect its most vulnerable new residents through initiatives like the ones suggested by Pollicy and the ones being spearheaded by the Raspberry Pi Foundation in order to ensure that Africa’s digital decade lives up to Google’s optimistic predictions for it.

This article Google’s $1 billion bet on Africa’s digital future is featured on Big Think.

If not for a busted bike chain, I would never have met Rick Doblin. In 2021, I was heading back to the INSIGHT Psychedelic Research Conference in Berlin trying to catch Johns Hopkins scientist Matthew Johnson’s talk. Fated with a twisted chain, I was late. Moments after staff ushered me upstairs to the balcony, Doblin sat beside me and started munching on a sandwich. We chatted about my book, which I’d just sent to be considered for publication. He leaned forward, eyes glimmering, and said, “What do you think’s gonna happen?” 

Now, three years later, I get to turn the question back on him.    

In August, the FDA rejected Lykos Therapeutics’ application to legalize MDMA-assisted therapy for PTSD. Lykos is the for-profit arm of MAPS (the Multidisciplinary Association for Psychedelic Studies), a non-profit founded by Doblin.

The rejection was a crushing blow to the organization and the field itself. MAPS has spent nearly two decades conducting clinical trials on MDMA-assisted therapy for PTSD and was slated to receive FDA approval after a successful Phase 3 study. However, the FDA rejected the application based on several factors outlined in an official statement, citing insufficient scientific evidence and flawed study design. 

Those concerns and others were raised in June when an independent FDA advisory panel voted against the treatment. The panel noted that Lykos failed to report data on “positive adverse effects,” like feelings of euphoria — information that’s relevant in assessing MDMA’s potential for abuse. It also considered a case of misconduct and the possibility that biases on the part of investigators, therapists, and patients might have swayed the reporting of harms and benefits, as described in May by a report from the Institute for Clinical and Economic Review.

“I absolutely agree that we need new and better treatments for PTSD,” said Paul Holtzheimer, a panelist on the advisory board and deputy director for research at the National Center for PTSD, who voted that the benefits of the MDMA therapy do not outweigh the risks. “However, I also note that premature introduction of a treatment can actually stifle development, stifle implementation and lead to premature adoption of treatments that are either not completely known to be safe, not fully effective or not being used at their optimal efficacy.”

In August, Doblin resigned from the board of Lykos, weeks after the company laid off 75% of its staff. As Lykos continues to work with the FDA on a path forward, the future of MDMA therapy for PTSD remains unclear, as does the question of whether it’s the drug, the therapy, or both in combination that may someday prove to be a safe and effective way to treat PTSD.

What is clear is that the FDA’s decision marks a major turning point for both Lykos and MAPS — and in determining which aspects of psychedelic treatment will be prioritized by stakeholders going forward. 

To blind or not to blind

To understand the FDA’s industry-rattling decision, it’s worth untangling a few threads in the story. The first is that MAPS applied not to legalize a drug but to get approval for its drug-assisted therapy — which involves patients taking MDMA and receiving Lykos’ particular brand of psychotherapy in an eight-hour session. 

“The way it’s been framed is that the FDA is against psychotherapy and that they don’t regulate psychotherapy; they just want to be about drugs,” Doblin tells Big Think. “That’s not the case. The FDA doesn’t regulate psychotherapy — that’s very true — but they understand that it’s not just about giving a drug.”

In a letter issued to Lykos, rejecting approval at the time, the FDA suggested they move forward by combining MDMA with what they called “evidence-based psychotherapy.” 

“Our therapy, I believe, is evidence-based,” Doblin says. “We’ve been studying it for twenty years. It’s produced great outcomes. We’ve studied it in research settings. Many of the companies are jumping on this idea of, ‘Oh yeah, let’s not do psychotherapy, let’s do psychological support.’ That’s not what’s in the best interest of the patients, but that might be what the pharma companies think will make them the most money.” 

But even if the therapy itself is sufficiently evidence-based, it’s less clear how to combine a drug like MDMA with it in a way that enhances, rather than diminishes, its effectiveness. Which leads to the other point of contention in this story: The FDA made a series of decisions over the course of the research process that may have prioritized patient safety at the cost of scientific rigor.  

When designing a clinical trial, the gold standard is to conduct a double-blind, randomized control trial where researchers try to reduce bias by ensuring that some trial participants don’t know they’re getting the drug in question. In many drug trials, that problem is addressed with an inactive placebo like a sugar pill, but seeing as psychedelics have such noticeable effects, psychedelic scientists often turn to active placebos instead. In other words: all participants receive an active drug, and can feel they’ve taken an active drug; they just don’t know which one it is. This approach, which allows researchers to study the effects of psychedelics with a higher degree of empiricism, is called “functional blinding.” By contrast, when that same group of trial participants simply receives an inactive placebo instead of an active drug, researchers run the risk of ”functional unblinding,” which occurs when participants can guess which treatment they received. Doblin says the FDA chose to take this risk, in a move that helped ensure participant safety but threatened the integrity of the trial design itself. 

“The FDA chose therapy with inactive placebo, and that kind of got lost. They betrayed their agreement with us by saying, ‘Oh, now we don’t believe these results because of the functional unblinding.’ We negotiated eight months in a special protocol assessment process with the FDA to come up with the design. The media has missed this whole real point that we understand functional unblinding. We discussed functional unblinding with the FDA at length, and that’s why they chose the inactive placebo.” 

Doblin, who wrote his dissertation on the blinding problem, says his solution was therapy with low-dose MDMA versus therapy with full-dose MDMA. “That is the best way to do a double-blind with psychedelics.” 

In sixteen years of phase-two pilot studies, MAPS compared inactive placebo with a wide range of doses: 25 milligrams, 30 milligrams, 40 milligrams, 50 milligrams, 75, 100, 125, and 150. 

“The challenge is to find the dose that’s high enough to cause some confusion with the full dose. A microdose doesn’t work, because you don’t really know you’ve got it, but not so high that you get so much therapeutic potential that then there’s no difference between the two groups, or it’s really hard to find a difference.” 

What they discovered was that the lower doses — 25, 30, 40, 50 milligrams — could produce some confusion over whether people are getting the full dose or not. But here’s the real kicker: they had an “anti-therapeutic” effect. 

“People who got therapy with low doses of MDMA still got better, but when you compare it to therapy with inactive placebo, people got better with the inactive placebo than with low doses of MDMA,” Doblin explains. “The reason is that the MDMA activated them, but it didn’t have enough of the fear reduction. These are the worst memories of people’s lives, and when they confront them in therapy, if you’re activated in a therapeutic setting, it can be very difficult.” 

That’s why, he says, with standard therapies — prolonged exposure and cognitive processing therapies, the two main therapies used in the VA for PTSD — there’s a dropout rate of roughly 50%. “Half the people can’t stand the therapy because the therapy says, ‘Talk about your trauma over and over and over’ and it’s difficult for people to do.” 

What they discovered during the research process, to their surprise, is that while the lower doses activated people and made them uncomfortable, the dose that was therapeutic was lower than they thought: 75 milligrams. For functional blinding purposes, that made it virtually impossible to find a sweet spot in dosage between a dose that causes confusion but doesn’t hurt the therapy and a dose that suddenly helps the therapy.  

“So we went to the FDA and we said, ‘We will give you blinding if you want blinding, but you’re going to make our job easier to find the difference between the two groups.’” 

In their official rejection statement, Doblin says, functional unblinding was cited as one factor influencing their decision.  

Somewhat ironically, the baseline recovery rates in the control group — without MDMA — were remarkable: 47.6% of chronic PTSD patients no longer met the criteria for the disorder at the two-month follow-up (compared to 71.2% who received MDMA and psychotherapy). These rates generally aligned with the first phase-three study, where one-third of patients with severe PTSD no longer met the criteria for the disorder after receiving psychotherapy but no MDMA.

“We had very minimal dropout rates. The [process involves] eight-hour therapy sessions […] When you have a one-hour therapy meeting once a week for, let’s say, a whole year, and the first 10 minutes you settle in. The next 20 minutes, you’re kind of getting deeper. Thirty minutes and then you’re starting to think, “Oh, my time’s almost up. Now I gotta collect myself and get ready to walk out the door, and I gotta focus on my traumas.” But when you have an eight-hour session with music, with two people to help you go inward, that can be incredibly effective, because people can choose when they’re ready to do the work with trauma, when they’re talking about other memories, when they’re talking about other issues. The essence of our therapeutic method is to support what’s emerging. It’s non-directive and inner-directed.” 

Had the FDA approved Lykos’ application, some might conclude there’s something special about the drug that is enhanced by the therapy. Others might think it’s the therapy that gets enhanced by the drug. In the aftermath of rejection, however, we’re left with another thread in a slowly unspooling truth about what makes psychedelics so hard to study. 

Inner-healer-assisted therapy

When I ask Doblin about mechanisms of action — how MDMA-assisted therapy works — his initial response is pragmatic.

“When you try to work with the FDA, you have to prove safety and efficacy, but you don’t have to prove any mechanism of action. Consequently, we have left the physiological mechanism of action to other scientists.” 

With a wry smile, he adds: “You hear about translational neuroscience. What I would say is that nothing is translated from neuroscience to help the clinician.”

In recent years, researchers have found that MDMA reduces activity in the amygdala, increases activity in the frontal cortex, and increases connectivity between the amygdala and the hippocampus, where memories are put into long-term storage. Oxytocin and serotonin work together synergistically to enhance neuroplasticity. 

“Those are good explanations, but none of that is really filtered down to how you do the therapy differently,” Doblin says, noting the thousands of years of psychedelic healing ceremonies preceding brain scans.   

Doing the therapy differently, however, is the exact platform on which MAPS has sold itself. 

“The main thing we’ve learned is that there’s this inner wisdom, this inner kind of logic to what’s emerging. It could be they’re moving into terror, they’re moving into shaking and just reenacting, reliving in their mind the trauma experience, or they’re talking about something in childhood, or they’re talking about happy memories. Whatever it is, the therapists don’t have an agenda. They support what’s emerging. We’ve called this the ‘inner healing intelligence.’ We know our body has this, that it heals itself in this quest to wholeness below our level of conscious awareness.” 

“Inner healing intelligence” — which MAPS’ treatment manual defines as “a person’s innate capacity to heal the wounds of trauma” — seems intuitive enough. After all, cuts and scrapes heal on their own without direct intervention from the mind, so couldn’t your mind “get out of its own way,” through ego dissolution or some similar mechanism, to help facilitate emotional healing?

MAPS’ treatment strategy is based on the assumption that it can. 

“Therapists should trust that any fear, memories, etc. that keep coming up are doing so to be healed, to be more fully understood, and that the participant’s psyche/inner healer knows when the best time is for this to occur,” states MAPS’ treatment manual. “This very process of surrendering ego/directed functions to self/inner healer may be the method of therapeutic action that is so hard to come by without the help of MDMA.”

Scientists have not empirically corroborated the existence of an “inner healer” or a “self” that seeks “wholeness” in the way MAPS describes it. Some also question the efficacy and safety of the non-directive therapeutic approach. Neşe Devenot, a senior lecturer at Johns Hopkins who specializes in psychedelic bioethics and submitted a citizen petition to the FDA arguing against Lykos, wrote that assuming there’s an “inner healer” guiding the processing of trauma may prime “both clients and therapists to associate distress with healing, even in situations where actual harm is occurring,” and that it risks encouraging study participants to “interpret any outcome as successful.”

Doblin is aware of such criticisms within the science community.

“‘Oh, this inner healing intelligence. It’s woo, woo. It’s not scientific. It’s, you know, it’s terrible, it’s scary. You’re smuggling religion into this and stuff like that.’” 

“I was just with a scientist from Johns Hopkins, and he said I should stop using the word ‘inner healing intelligence’ because people don’t know what we mean. None of that is true. All it is is just to say that we support what’s emerging, whatever it is: If it’s terrifying, if it’s pleasant, if it’s just neutral, whatever it happens to be. We let people come around to the trauma at their own pace, and people will always eventually get to it — because that’s what their problem is. But they might not get to it in a straight line, and that’s why the cognitive processing therapy and the prolonged exposure are so re-traumatizing. You say, ‘We’ve got an hour meeting. Tell me about your trauma.’ It’s just a different kind of approach.” 

The idea of an inner healer phenomenon has caught the interest of certain figures within the psychedelic science community. In a post last year on X, the neuroscientist Robin Carhart-Harris wrote: “Can someone quote me something from a book or in academic literature referring to the inner healing mechanisms of psychedelic therapy?” Several months later, he posted: “A small step on a brave journey to consider & examine the validity of the ‘inner healer’ phenomenon in context of psychedelics,” sharing his new paper on the topic, titled, “Psychedelics and the inner healer: myth or mechanism?” Carhart-Harris wrote that he “does not see it as a mystical/supernatural phenomenon at all.”

Doblin notes that proponents of the inner healing approach argue that it “helps people feel that they can trust what’s happening, even if it’s leading them into fear and terror,” adding that it’s a gentle salve for moments during a trip when someone might be thinking, “I’m going to lose control, I’m going to be crazy.” 

“Difficult is not the same as bad in terms of psychedelic experiences,” Doblin tells me. “What turns something difficult into something bad is resistance. So I think, again, part of the mechanism of action is that you open up to it and you descend into it. If you are fully open to it–life is constant change–something will change.” 

The future of psychedelic therapy

In October, Lykos issued a press release after a meeting with the FDA about the path forward. 

“They have to do another Phase 3 study, review the videotapes, do various things,” Doblin says. “What we are now currently thinking is probably three years from now, they’ll come to an agreement and FDA approval will be very likely.” 

When I ask Doblin how Lykos and MAPS will navigate forward from here, he expresses a few concerns. 

“The switch I worry about [is moving away] from maximizing therapeutic outcomes — which I believe has been essential for changing the whole culture, changing the ecosystem, getting people enthused about psychedelics, getting MDMA research throughout the VA — to now the pharma company way forward. And this is true for other psychedelic pharma companies too — just looking at what’s the minimal viable product, what’s the lowest cost.” 

Questions remain: What type of psychotherapy, if any, might be paired with MDMA if the FDA does approve MDMA therapy? Does psychedelic therapy even need therapists? Or will trained sitters without a therapy background suffice?

“Lykos, in particular, is all about the FDA. Lykos will eventually succeed, I believe.”  

MAPS the nonprofit has plans of its own. Doblin tells me he just got off a two-hour call with a woman from Rwanda who wants to bring MDMA to Rwanda. MAPS’ itinerary of global humanitarian projects so far includes work in Somaliland, Denmark, Bosnia, and an upcoming training in Ukraine in the first quarter of 2025. 

“The pharma company is going where the suffering is and where there’s money, and MAPS the nonprofit is going where there’s suffering but where there’s no money.” 

The organization also plans to start focusing on group therapy, both to reduce costs and to see whether “those groups can be supportive of each other in the integration process, and maybe they’ll get results comparable to the outstanding results that we get from working with individuals.”

In October, Lykos published a statement saying it had a “productive” meeting with the FDA regarding the next steps for gaining approval for its treatment, and that the company “will continue to work with the FDA on finalizing a plan.”

This article Inside the battle for FDA approval of MDMA therapy is featured on Big Think.

Among the many casualties of the coronavirus pandemic are the poorest of the global poor. Not only are they the ones who are usually hit hardest by the virus, but they’ve also borne the brunt of virus-induced wreckage in the world’s economies.

Before the pandemic hit the past four decades had seen global poverty rates plummet to less than a quarter of what they were in 1981. But that trajectory halted abruptly in the infamous year 2020, which kicked nearly 100 million people back into “extreme poverty,” defined as living on less than $1.90 per day. Nobody knows when or if they will recover.

Even in the United States, where the effects have been cushioned by trillions of dollars in federal pandemic relief, the economic recovery is leaving behind women, Black people and other historically disadvantaged groups.

The good news, though, is that the fight against poverty and inequality is resuming with tools and programs that are more effective than ever before, thanks to a quiet revolt against business as usual that is increasingly being embraced by researchers, activists and government officials.

Known as the evidence-based policy movement, this shift in people’s thinking dates back more than 100 years. But it really took off when tax revenues cratered after the Great Recession of 2008, says Sara Dube, director of the Pew Results First initiative. “States were searching for any sort of tangible tools or processes that could help them make the most of very limited resources,” says Dube. And the evidence-based approach offered them just that. Instead of just fielding more and more social support programs designed the standard way — via “good intentions, inertia, hunches, partisan politics, and personal relationships,” as a 2013 op-ed put it — the idea is to first apply the methods of science to see if the programs actually work.

Recent decades have seen a dramatic decline in the absolute number of people living in extreme poverty, defined as living on less than class=”wp-element-caption”.90 per day, in nearly all regions of the world and despite continued population growth. Before the Covid-19 pandemic, as seen in this projection from 2019, that decline was expected to continue for many areas at least until 2030.

A prime example is the once popular “scared straight” intervention, which tried to steer at-risk juveniles away from crime by showing them how awful prison could be. It seems like common sense. But whenever this approach is subjected to the same kind of randomized controlled trials used for testing drugs and other medical treatments — that is, whenever researchers compare a random selection of subjects who get the intervention to a control group of similar individuals who do not — the results are unequivocal: Scared-straight programs are worse than useless. If anything, they lead to an increase in criminal behavior. Many (though not all) jurisdictions have now abandoned them.

The use of randomized controlled trials to assess social programs, honored by the 2019 Nobel Prize in economics, is one of the most powerful methods in the evidence-based toolkit. Another, honored by the 2021 economics Nobel, is exploiting “natural experiments” in which similar groups do or don’t get a treatment by happenstance. An oft-quoted example is when New Jersey raised its minimum wage in 1992 and neighboring Pennsylvania did not: A comparison of the two states’ subsequent employment growth showed that higher labor costs did not depress hiring, despite the dire predictions of many traditional economists.

Still another method is to mine the troves of “administrative data” that government agencies routinely gather about the people they serve. These data streams, which include school attendance records, court documents, housing inspection results and much more, can be aggregated into a rich portrait of how people live, how they are faring and how various programs affect them. Congress gave a ringing endorsement to this kind of analysis when it passed the Foundations for Evidence-Based Policymaking Act of 2018, which requires each federal agency to make its administrative data publicly available (with appropriate privacy safeguards), and to develop a formal plan for evidence-based program evaluation.

Of course, none of this is going to solve the problem of poverty anytime soon; it’s far too complex and entrenched for quick solutions. But by putting the focus on data, as opposed to ideology and guesswork, the evidence-based policy movement is providing a much surer path toward that goal.

To see how this works in practice, consider what evidence can tell us about the newly prominent (and highly contentious) notion of giving everyone a basic income — no strings attached.

Compared to projections from 2017, the last year with official global estimates, the Covid-19 pandemic knocked an estimated 97 million people back below the extreme poverty line.

Evaluating universal basic income efforts

To the extent that poverty is just about being broke, giving people cash is the obvious solution. And of course, many developed nations do that already, through food stamps, unemployment benefits, child tax credits, welfare payments and a multitude of other cash transfers. What’s different about the basic income idea is the “no strings attached” part: The payments would go to everyone, with no work requirements and no questions asked about how rich or poor a recipient might be. (That’s the theory at least — many proposed programs do in fact include restrictions on who would receive the money.)

This is actually a very old idea, dating back at least to Thomas More’s 1516 fantasy Utopia. It helped shape the Social Security system during the 1930s, when Depression-era agitation to give the elderly a universal basic income was widespread. And it’s enjoying quite the resurgence in the United States these days — most notably during the lead-up to the 2020 Democratic presidential primaries, when candidate/tech entrepreneur Andrew Yang drew outsized attention with his proposal to tackle poverty and technological unemployment by giving every adult American $1,000 per month. But Yang was not operating in a vacuum. In 2019, for example, Stockton, California, had already started sending $500 per month to a selection of households in the city’s low-income areas, making it one of the first modern US cities to start a randomized controlled trial of a basic income. And many other cities have followed suit — especially during the economic chaos of the pandemic.

Perhaps not surprisingly, advocates on the political left see the basic income as a way to restore dignity and stability to chaotic lives, not to mention helping poor people pay for basic needs like food and shelter. But the idea has also drawn substantial support from conservatives — including such libertarian icons as the late economists Milton Friedman and Friedrich Hayek. If nothing else, goes the libertarian argument, a basic income could largely do away with the current welfare state: Instead of asking governments to micromanage dozens or hundreds of individual social programs, each with its own rules and bureaucracy, just give people the money to spend as they wish on the open market — and let them take responsibility for their choices.

Starting in 2019, the Stockton Economic Empowerment Demonstration (SEED) in Stockton, California, gave 125 residents 0 per month for two years, no strings attached. They mostly spent their money on necessities, not “temptation goods” such as alcohol or tobacco.

To many others, however, especially (but not exclusively) on the right, the basic-income idea is anathema. Quite aside from the cost, which the government would have find a way to fund, “free” money is widely seen as discouraging work, rewarding bad choices, creating a culture of dependency and being an open invitation for recipients to squander every cent they get. As Republican Senator Chuck Grassley of Iowa put it, referring to the 2017 tax cut, the cut “recognizes the people that are investing, as opposed to those that are just spending every darn penny they have, whether it’s on booze or women or movies.”

From the evidence-based policy perspective, however, the impact of a basic income is a question to be settled by data — a stance that has resulted in an extensive research literature. In 2017, for example, economist Ioana Marinescu reviewed a number of randomized trials and natural experiments that have mimicked aspects of a basic income in the North American context. Marinescu, now at the University of Pennsylvania, included the 1970s-era US and Canadian trials of a “negative income tax” that returned money to people with earnings below a certain threshold; the Alaska Permanent Fund that has shared oil revenues with state residents since the early 1980s; a casino dividend that’s been distributed to members of the Eastern Band of Cherokees since 1997; and even lottery winners. Writing in the 2019 Annual Review of Economics, researchers likewise looked at basic-income-like initiatives from the developing world — among them a nationwide cash transfer launched by Iran after the government halted its food and fuel subsidies in 2011. And in 2020, the city of Stockton released an interim report on the first year of its basic-income experiment, written with independent researchers from the Universities of Tennessee and Pennsylvania.

What’s striking about these studies and many others is how consistent the findings are. Sure, says Paul Niehaus, an economist at the University of California, San Diego, and a coauthor of the 2019 review, if you’re in a cherry-picking mood “you can easily find stories to confirm your view of the world.” There will always be that one individual who does indeed squander all his money on “temptation goods” like alcohol, drugs or tobacco. But when you aggregate the data over all of the study’s participants, says Niehaus, such individuals tend to be the exception, not the rule. Even in the most deprived populations, he says, he and his colleagues find that recipients still spend much of their money on things like food, shelter and education — or even, in some cases, starting new businesses.

Some studies also show that recipients are at least somewhat more likely to have jobs than their counterparts in the control group — possibly because having a more stable financial situation makes it easier for them to go out and look for work. (Among the few examples to show a decline in employment rates were the 50-year-old negative income tax experiments. But even there, according to Marinescu, the drop barely had statistical significance.) And recipients often report substantial improvements in quality-of-life measures like nutrition, health and educational attainment.

That said, adds Niehaus, few if any of the experiments to date have been a truly rigorous test of the basic-income idea, mainly because of their limited scope. In some cases, he says, “they’re small-scale in the sense that they’re enrolling individual people and not entire communities.” So would more new stores (and jobs) open up if everyone in the neighborhood got paid? In other cases “they’re small-scale in the sense that they’re giving people money for a year or two, but not making a long-term commitment,” says Niehaus. So would people’s spending habits be any different if they knew their basic needs would be met for the next decade, and not just for the next year?

A basic-income experiment carried out by Finland in 2017-18 found that recipients felt more economically secure than a control group. Scores for days of employment, life satisfaction and self-reported depression were also slightly better for recipients than for controls.

Since 2017, Niehaus has been part of a team of researchers trying to answer such questions with the largest and longest trial of the basic-income idea yet. Encompassing 14,474 households in Kenya, the experiment has randomly assigned 295 villages either to a control group that gets no payments; a lump-sum group in which every adult villager gets a one-time payment of about $500; a short-term group in which every adult gets 75¢ per day for two years (enough to meet basic needs in Kenya); and a long-term group in which the same daily payments will last for 12 years.

He and his colleagues are already analyzing some of the interim results, says Niehaus. When the Covid-19 pandemic hit, for example, they found that basic-income recipients generally experienced less hunger than control-group members, and had less sickness in their households.

Niehaus and his group are also planning their post-pandemic return to the field: “A key point here is that we care about the impacts on behavior now of expecting to receive transfers in the future,” he points out, “so we don’t need to wait 12 years to learn about some of the most important questions.”

A need to address systemic barriers in poverty

Even when those answers are in, however, that still leaves a host of questions that data alone can’t satisfy. How would a basic income program be funded, for example? Would it be universal, going to everyone, or targeted at the neediest? Would it supplement existing anti-poverty programs, or replace them?

And even if you somehow reached enough consensus on those questions to get a basic income enacted, that still wouldn’t be enough. “Poverty isn’t just about the lack of money, it’s about the lack of opportunity,” says Carrie Cihak, who is both a research affiliate at the Center for Advanced Study in the Behavioral Sciences at Stanford University, and policy director for King County Metro, the public transit agency for a swath of Washington state that stretches from Seattle and its suburbs to the Cascade range.

Those opportunity shortfalls comprise a very long list of barriers that include lack of access to childcare, health care, good food, decent housing and quality education; mental health and addiction issues; the impact of an arrest record on job prospects — and yes, the legacy of past and present racism.

But a particularly frequent pain point is transportation, says Cihak, which is why her agency is interested: “Human service agencies are constantly scraping and scrambling to figure out ways to get their clients to something,” from court appointments to health visits to a community college class to the kids’ drop-off at preschool. “It is a constant struggle for people,” she says.

Obviously, says Cihak, governments at every level are already trying to address all these barriers with any number of social support programs, from transit subsidies to job training initiatives. And, as documented in an extensive database of studies maintained by the advocacy group Results for America, evidence-minded researchers and agencies are evaluating such programs as fast as they can.

Yet the systemic-barrier problem remains a research frontier for the evidence-based policy movement, says Cihak. It’s like the difference between testing a new drug and reforming the health care industry: Randomized controlled trials and all the rest are great at evaluating the effect of specific interventions on specific populations under specific conditions — but not for fixing issues that are so interlinked and intertwined that they can’t be solved in isolation.

“Poverty isn’t just about the lack of money, it’s about the lack of opportunity.”Carrie Cihak

For problems like this, says Cihak, “evidence-building takes time. It isn’t about one study but is a continuous learning process” — something that’s always been tough for social service agencies constrained by funding, pressure to get services out the door quickly and institutional rules and practices that may not always serve the people they’re trying to help.

Again, though, the good news is that these agencies seem to be undergoing a noticeable culture change. Thanks in part to the evidence-based policy movement, says Jake Bowers, a political scientist at the University of Illinois who coauthored an assessment of the movement in the 2019 Annual Review of Political Science, “a growing number of people who are in charge of providing services from different governments are willing to try out new ideas.” They are also approaching the social service development process with a new kind of humility, says Bowers: Instead of developing an anti-poverty program in isolation and trying to sell it as something that is absolutely, definitely, guaranteed to work, they’ve begun to treat the process as a chance for testing, being wrong and learning.

There are even signs of change in the political sphere. “The hope is that good science doesn’t just produce good policy, it produces good politics,” says Paul Testa, a Brown University political scientist and Bowers’s coauthor. So even when competing sides violently disagree on principle, he says, “if they could agree on a process to generate evidence of what works best, that would help reach consensus.”

Granted, that seems like a pretty big “if” these days. But maybe not. The aforementioned Evidence Act of 2018 managed to sail unscathed through a Congress mired in bitter partisan gridlock, after all. It passed both houses with near-unanimous bipartisan support. And it was signed into law without a murmur. By President Donald Trump.

“There’s understandably a lot of cynicism out there about how policy is made,” says Richard Hendra, data chief for MDRC, a nonprofit that’s been doing evidence-based policy research since its founding in 1974 as the Manpower Demonstration Research Corporation. But after working with both parties, he says, “I’ve been impressed how much people do listen to evidence.”

This article Can science solve the poverty problem? is featured on Big Think.

It’s been proposed that anthropogenic climate change is simply a problem of waste mismanagement. In comparing carbon to sewage, scholars have contended that tools and technologies must be developed to engineer carbon out of the atmosphere and into the sinks within the geosphere and biosphere.

While this sounds like an apocalyptic techno-fix for some, the IPCC concluded that there is a significant need for negative emissions technologies on an industrial scale (a magnitude of 1.25 billion tons a year) to remain under the 1.5°C global temperature target set by the 2015 Paris Climate Accord. As Myles Allen, director of the Oxford Net Zero Group (ONZ) and leading geophysicist within the Intergovernmental Panel on Climate Change (IPCC), once declared, “We must stop fossil fuels causing global warming before the world stops using fossil fuels.” It’s no longer the stuff of science fiction—fossil fuel burning has unlocked a reservoir of carbon that’s far in excess of the ability of nature to sequester.

This means forms of human-enhanced carbon capture that will increase the planet’s capacity to “recycle” carbon—without raising the global surface temperature—are on the table for immediate discussion and implementation. But while it appears essential to unlock carbon stores that can hold carbon on timescales equivalent to fossil fuels, this solution obscures the reparative function of these technologies; it can’t simply be about restoring a “carbon budget.” How can we decarbonize in a way that moves toward liberatory futures for communities that have faced disproportionate harm from the anthropogenic climate crisis?

Many equity issues surround the concept of reaching “Net Zero,” which the ONZ defines as “the state in which the greenhouse gases going into the atmosphere are balanced by removals out of the atmosphere.” Over the past two centuries, overdeveloped countries, which consist of former and current colonial powers, have historically emitted a larger share of fossil-based emissions and disproportionately contributed to deforestation for natural resource extraction as estimates approximate that the US still ranks as the highest emitter per capita, responsible for 20 percent of cumulative carbon emissions over the past two centuries.

As underdeveloped countries continue to industrialize in this century, emitting a fraction of fossil fuel and land use change emissions compared to their counterparts, they’re penalized by intergovernmental organizations now looking to enforce strict climate targets. Climate targets without regard for a global history of uneven development have upheld the apolitical, ahistorical framing often associated with the anthropogenic climate crisis. Here’s where a difference between luxury and survival emissions emerges, even though they’re homogenized in international carbon litigation, as emissions resulting from subsistence activities are judged at the same weight as those generated from high-emitting cars and air travel.

Negative emissions technologies, nature-based or otherwise, have the potential to create a fair allocation of carbon to make room for the communities yet to industrialize while also mitigating anthropogenic climate change that’s still avoidable. But some corporations and state actors consider the atmosphere a new commodity frontier, shaping how carbon removal technology is developed. Climate scholars Adam G. Bumpus and Diana M. Liverman argue that corporations have been more focused on the opportunity to reduce costs by deferring carbon removal responsibility elsewhere through carbon markets and offsetting rather than decarbonizing practices within their own organizations. The carbon market and offsetting system have created “carbon cowboys” and perpetuated forms of neo-colonialism, land-grabbing, and further undermining of the agency and sovereignty of vulnerable communities around the world.

Yet not all forms of carbon removal are created equal. Perhaps, in dreaming of a new future for sustainable development, communities can center the development of carbon infrastructure that is life-giving, abundant, and alimentary while returning lands to and prioritizing well-being for communities that have faced historic injustices. Land-oriented carbon removal, such as restoring wetlands, promoting holistic nature recovery strategies, and utilizing agroecological growing techniques, can put the role of carbon removal back into the hands of land stewards everywhere.

Borrowing from scholarship on the principle of “extended producer responsibility (EPR),” Allen and co-authors Stuart Jenkins, Margriet Kuijper, Hugh Helferty, and Cécile Girardin propose that we hold producers accountable by requiring them “to provide or contribute to the prevention and management of the resulting waste” throughout the fossil fuel lifecycle. Environmental social sciences scholars Andrew Bergman, Holly Jean Buck, Olúfẹ́mi O. Táíwò, and Toly Rinberg suggest that even industrial-scale carbon removal technologies can be made more democratic, decentralized, and convivial to ensure that all can have a say (and a part!) in developing carbon removal. This doesn’t just look like diversifying existing carbon supply chains nor giving the state ultimate control of carbon development. Instead, they challenge us to truly change how we think about the ownership of carbon, the normalization of carbon-heavy consumption patterns, and the responsibility of carbon production. What could it look like to mobilize and compensate communities for supporting carbon drawdown efforts, ensuring they have sovereignty and agency over their participation in carbon removal development?

Pursuing carbon removal technologies within an agenda of transformation and abundance brings us to a critical crossroads in addressing the environmental crisis and historical (and current) injustices. As the imperative to mitigate anthropogenic climate change intensifies, the discourse surrounding carbon removal often revolves around technological solutions and achieving net zero. However, the essence of this pursuit goes beyond mere atmospheric carbon balance; it encompasses rectifying imbalances and embedding equity within the fabric of these solutions, creating wholly new futures for carbon.

This article Not all forms of carbon removal are created equal is featured on Big Think.

Have you met Gary Marcus? If not, let me introduce you. A cognitive scientist, Marcus serves as professor emeritus of psychology and neural science at New York University. He has authored several popular books, offered testimony during a Senate hearing on AI, and co-founded Geometric Intelligence, an NYU-incubated AI startup later acquired by Uber. When not at work, he is busy raising two young children and Henrietta, his pet chicken.

However, Marcus is best known today as AI’s loudest critic — or denouncer or nitpicker. It depends on who you ask.

Science fiction writer Kim Stanley Robinson called him “one of our few indispensable public intellectuals.” AI heavyweights Yann LeCun, chief AI scientist at Meta, and Geoffry Hinton, the “Godfather of AI,” have publicly debated Marcus. Meanwhile, Sam Altman, OpenAI’s CEO, has been known to throw occasional shade at some vague “mediocre deep learning skeptic.” Other AI evangelists have been less elusive. They argue that Marcus isn’t just wrong but actively gaslighting the public over the imminent AI singularity.

Everything I wrote about Marcus is true except for one detail: his feather baby, Henrietta. She was created out of whole cloth by ChatGPT when Tim Spalding prompted the popular AI chatbot to write Marcus’ biography and mention his pet.

Henrietta is a “hallucination,” a fanciful term for the fabrications and factual inaccuracies produced by generative AI. Researchers don’t have a definitive answer to how often these AIs hallucinate — estimates range from 3–27% of the time — and because the intermediate layers of these systems’ decision-making are impenetrable, no one can say why ChatGPT fashioned Marcus a poultry papa.

Marcus speculates the AI scraped his Wikipedia page alongside data about the children’s book Henrietta Gets a Nest, illustrated by Gary Oswalt. Because large language models (LLMs) like ChatGPT don’t use reason but eke out statistical associations between words across vast amounts of data, it may have glommed together these two Garys with some general info about pets. When prompted to discuss his supposed animal inspiration, ChatGPT hallucinated and voila! Kentucky counterfeit chicken.

Now, a phantom pet may seem a trifling matter, and in the grand scheme, it is. However, generative AIs don’t recognize the difference between important and inconsequential tasks. They are designed to provide the best answer, image, or whatever to the user’s prompt, and they fail at this, Marcus argues, more often than any user should be comfortable with.

“I always think of this expression from the military: ‘Frequently wrong, never in doubt,’” he said during our chat. “Large language models say everything with absolute authority, yet much of what they say is not true, and they don’t let you know that. We should be hoping for a better form of AI.”

After our conversation, I came to think of Henrietta as more than a humorous hallucination. She became a handy representation of the two major concerns Marcus has with the current AI environment: the inadequacies of the tech and the moral shortcomings of the industry. They’re troubles that Marcus believes we should all be more concerned about as these systems, and the companies that own them, gain more power in our lives and the societies we live in.

All rhyme, no reason

Let’s start with the technical shortcomings. In 2022, Marcus wrote an article titled “Deep Learning Is Hitting a Wall.” Many ridiculed the premise at the time, pointing to the field’s achievements and the promise of more to come. As if taking its cue, OpenAI released ChatGPT-4 the very next year, blowing everyone away with the chatbot’s capabilities and seemingly putting Marcus’ claim to bed.

"Deep Learning is hitting a Wall" — Gary Marcus pic.twitter.com/PQ4RiLCbIu

— Joscha Bach (@Plinz) March 11, 2022

But his argument wasn’t that deep learning — the method used to train generative AIs — wouldn’t see advancements in dimensions such as functionality or processing power. The wall in question consisted of those cognitive capabilities shared by truly intelligent agents. These include things like planning, reasoning, reliability, an understanding of how the world works, and recognizing the difference between facts and misinformation.

“None of those problems have been solved,” Marcus says. “Despite much more money going to the field than before — probably ten or a hundred times more — better chips, more data, more competitors in the game, more people taking shots on the goal, nothing is really much better than GPT-4.”

Over on his Substack, Marcus has made chronicling the ways contemporary AIs fail to act intelligently something of a hobby. And his examples are legion (for they are many).

Here’s one: In 2022 Meta’s Galactica AI weirdly claimed Elon Musk died in a car collision on March 18, 2018 — despite four years of headlines, business deals, and many, many tweets to the contrary. Again, no one can say why the model made this error. One ventured guess is that the system merged Musk’s name with information about the 2018 Tesla crash that killed Walter Huang, a former Apple engineer. After all, Musk’s name is statistically more likely to appear alongside the word Tesla than Huang’s.

Speaking of Musk, his company’s AI chatbot, Grok 2, has been caught spinning fake headlines and spreading misinformation, as well. Earlier this year, Marcus asked it to produce a picture of Italy’s current prime minister, Giorgia Meloni. The results? Grok delivered four images. All of them were men. Two of them were of US President Donald Trump.

Face it, @elonmusk: Grok 2 can be 𝙪𝙣𝙗𝙚𝙡𝙞𝙚𝙫𝙖𝙗𝙡𝙮 𝙨𝙩𝙪𝙥𝙞𝙙.

Consider “Please make a drawing of the current prime minister of Italy; the actual person in office right now.”

In 4 tries, despite ample available training data, not even the gender is right. pic.twitter.com/a7sPBjeJ1Q

— Gary Marcus (@GaryMarcus) August 16, 2024

Google recently tried to weed out this apparent bias from its Gemini AI. The company’s engineers trained the model so that if a user asked for a picture of, say, “an American woman,” Gemini wouldn’t produce images of only white women but a more representative sample. It was a well-intentioned effort, but one the AI took too far. It generated all sorts of historically inaccurate images, including those of black men and Asian women sporting the Iron Cross as Nazi-era German soldiers.

“A genuinely intelligent system should be able to do a sanity check,” Marcus says. “It should be trivial, if we were actually on the right track, [for AI] to check facts against Wikipedia.”

famous river crossing puzzle, and it’ll get all three across the river uneaten. But as Marcus notes, the system isn’t using reason to solve it. It instead relies on pattern matching to see how others have previously solved identical or similar problems — like autocomplete on steroids.

Because of this, large language models have difficulty with inference, mathematics, and novel conundrums. They even get tripped up when users change a word or two in well-documented problems. Modify the river crossing puzzle slightly, and LLMs tend to spit back nonsense because they’re stuck on the pattern of the original. (OpenAI’s o1 model has improved reasoning in some respects, but Marcus argues it is still flawed and limited.)

Reliability problems even crop up when models only need to cite preexisting information. Last year, a lawyer turned in a 10-page brief written by ChatGPT. The brief cited dozens of court decisions the AI claimed were relevant to the lawsuit, but upon closer inspection by the judge and opposing counsel, it was discovered that several were made up. Similarly, various AIs have advised the more culinary-minded among us to roast potatoes with mosquito repellent, mix glue with cheese for a tastier pizza, and saute Sarcosphaera mushrooms with onions and butter (to really bring out those umami and arsenic flavors, one supposes).

Gary's out here dunking on GPT 4o again, and I'm here for it, but in my experience it isn't stupid as much as overfit, at least for this particular puzzle. https://t.co/ZwN0y9bygJ pic.twitter.com/EbOxRzsFCK

— Sam Charrington (@samcharrington) May 15, 2024

“If you want to believe in this technology, you want to think you’re in this magical moment when the world changed, then you notice the things it gets correct,” Marcus says. “If you’re a scientist, you also think about what goes wrong. What does it mean if it is 70% correct? Should I use it?’”

Mind the gullibility gap

Of course, people make mistakes, too, and any technology is only as good as the user’s understanding of it and its limitations. AI is no exception. 

Yet despite this lack of reliability and trustworthiness, large language models still present their results with authority and a veneer of intelligence. This presentation style preys on the vulnerabilities of our mental makeup, leading users to fall into what Marcus calls the “gullibility gap.” That is: Our tendency to ascribe intelligence to non-intelligent things simply because they feel that way. It’s comparable to pareidolia, a phenomenon in which people see patterns in random visual stimuli such as a rabbit on the moon or the Virgin Mary in a burnt grilled cheese.

“I often hear people say things like, ‘GPT-4 must be a step toward artificial general intelligence,’ and I say not necessarily. You can get intelligent-looking behavior for a bunch of different reasons, including just repeating things you’ve memorized from a larger database,” Marcus says.

The pull of the gullibility gap can be strong, even for people who should know better. For instance, in 2023 the tech website CNET quietly published several articles penned by an LLM. The articles were well-written, and the editors, who would have known about hallucinations, were supposed to fact-check them before publication. However, the number of “very dumb errors” found by readers revealed they did the opposite. They put their trust in the AI and set their brains on autopilot. 

For Marcus, it has been unsettling to watch how easily smart people — from writers to lawyers — have fallen for the AI hype and careened right into the deep learning wall.

The moral AI ground

Let’s return to Henrietta and Marcus’ second concern: what he sees as the moral shortcomings of the industry. Again, there’s no way to know for sure whether ChatGPT borrowed from Henrietta Gets a Nest to confabulate Marcus’ poultry pet. Even if that’s not the case, many other instances reveal that tech companies bundled large swathes of copyrighted material from the internet to train their AI — and without asking for permission first.

The New York Times lawsuit against OpenAI houses several examples of GPT-4 responses that are near verbatim recreations of Times stories. Similarly, Marcus and Reid Southen, a film industry concept artist, have documented many cases of Midjourney and Dall-E 3 producing images containing copyrighted characters from popular movies and video games.

If Claudine Gay resigned amid plagiarism charges, where does that leave OpenAI and Midjourney, which regularly produce near verbatim replicas of trademark characters, at the slightest prompting, never ever providing attribution? pic.twitter.com/J8qVmqOy1o

— Gary Marcus (@GaryMarcus) January 2, 2024

It remains an open legal question whether AI training data must be licensed beforehand or whether the practice falls under fair use. Still, as Marcus and Southen note in their article, AI also presents a values question separate from the legal one: Should AI companies be allowed to use artists’ work without their consent or fair compensation? Because while the Times may have the clout and resources to challenge OpenAI, the same will not be true of the many small companies and individual artists whose works may have been hoovered into the algorithm.

Marcus has also been highly critical of the push against regulations. While many in Silicon Valley have expressed concerns over AI gone awry — even Altman has stated misgivings over “very subtle societal misalignments” — the industry has used its immense war chest to fight or rewrite the laws it finds stifling.

In his book, Taming Silicon Valley, Marcus cites a Corporate Europe Observatory report detailing the industry’s lobbying efforts to water down amendments to the EU AI Act. According to the report, before the law’s drafting period, the industry increased its lobby budget to €133 million, a 16% bump. It also gained outsized access to lawmakers. Of the “97 meetings held by senior Commission officials on AI in 2023, […] 84 were with industry and trade associations, twelve with civil society, and just one with academics or research institutes.”

“You can’t blame corporations for wanting to speak up, but the governments need to work harder to include independent scientists, ethicists, and other stakeholders from civil society,” Marcus writes.

At least the EU AI Act passed — even if tech companies’ compliance with the Draft Act has so far been low. Across the pond, Silicon Valley also lobbied against California’s AI safety bill and Biden’s AI executive order. The former was vetoed by California Governor Gavin Newsom, while the latter faces an uncertain future in light of the 2024 elections. Meanwhile, political gridlock has prevented the US Congress from taking meaningful steps.

Of course, not all regulation is good regulation, and arguments exist for where the laws mentioned above fall short. More general objections to AI regulation claim it will slow down advancements in the field, stifle competition with compliance costs, reduce investment and job creation, and impose massive burdens on small and open-source players — all without clear benefits.

For Marcus, such arguments aren’t convincing enough to eschew regulation entirely. They are merely additional factors to consider when crafting sensible regulations, and any sensible regulation must also be sure to avoid regulatory capture, keep the field open to up-and-coming companies, and, above all, protect people and their rights. 

“You can’t rush these technologies and ram them down people’s throats,” Marcus says. “We want AI to be astonishingly safe because everybody in the business realizes that bad things could happen, [and] things could get worse. The more power that we give these systems, the more something serious could go wrong.”

The lack of openness and transparency on the industry’s part extends to the many other ethical concerns Marcus has with the current AI landscape: security risks, the environmental impact, the inability to reason against the biases lurking in datasets, and the lack of guardrails against AI hypercharging misinformation campaigns that could short the market, promote online theft, and foment widespread political vitriol.

“If we use generative AI, and it’s stealing from artists and writers, it’s contributing to environmental harm, it’s contributing to racism, etc., are we part of the problem?” Marcus asks. “If you just buy the product and you don’t let [these companies] know that this behavior and these actions are not acceptable, they have no incentive to go the other way.”

Yet another brick in the wall

If you’ve read this far, you may also have gotten the idea that Marcus has earned his reputation as a nitpicker or denouncer or critic for criticism’s sake. At the very least, he’s anti-AI, right?

I apologize if that’s the impression because it’s definitely not the case. During our conversation, his enthusiasm for technology and its ability to broaden our conception of intelligence was evident. You could speak with him for hours about this stuff, and while he’ll certainly push back against ideas he finds lacking, it’s clear that he believes AI has the potential to benefit people and societies.

He praised David Baker, Demis Hassabis, and John Jumper’s Nobel Prize win for their work on the AlphaFold, calling it a “huge contribution to both chemistry and biology” and one of the “biggest contributions of AI to date.” He even admits that large language models can be useful when brainstorming, doing research prep, or helping coders — basically, anything where you only need “rough-ready results” and have the space to be mindful of the gullibility gap.

He isn’t a modern-day Ned Ludd riotously jailbreaking LLMs out of spite. As he said at the top, he wants us to have “a better form of AI,” and he worries that the breathless hype surrounding the current iteration of this technology has blinded us to its downsides and the far-reaching consequences of continuing the status quo. As he mentioned during our chat, “I’m a big fan of AI and part of the reason large language models upset me as a scientist is that the technology is so popular that it distracts from the other scientific work that we need to do.”

For instance, the massive financial and energy investments placed into deep learning make it difficult for alternative hypotheses to compete. Should scaling these neural networks with even more data, energy, and processing power not solve their technical shortcomings but instead lead to diminishing returns, then alternative approaches will be necessary to make them reliable and trustworthy.

Marcus contends that the most promising path forward is a hybrid approach, one that combines the benefits of neural networks with a neurosymbolic AI that uses the formal logic of traditional computer programs. But he adds:

“Even if I’m wrong, it’s safe to say that we have given an incredible amount of resources to the scaling hypothesis, and the hallucinations won’t go away. That’s just not working,” Marcus says. “If you were to do that in a traditional science context, people would say this question has been asked and answered. Why don’t you try a more innovative approach? Do something different!”

Similarly, if we don’t address the moral shortcomings of the status quo, Marcus fears they will ignite a “backlash against the entire field of AI.” Here, Marcus recommends users and citizens take more action directly. In Taming Silicon Valley, he offers an action plan that citizens should direct their representatives to pursue. The plan includes pushing for more transparency and better oversight; international AI governance; and better protections surrounding privacy and data rights.

“Maybe we will boycott products that aren’t addressing these problems. Maybe we will vote people out of office if they don’t take it seriously,” Marcus says. “This is going to change our world — whether we have an AI policy that gears AI to be for the many rather than the few.”

And as Henrietta suggests, those changes can get existentially weird. While ChatGPT certainly hallucinated the bird, the chatbot was also proven right in a fashion. Henrietta has served as an inspiration and insightful lens through which Marcus has contemplated the complexities of AI. He may never have owned a pet chicken, but she’s now an inseparable part of his story either way.

This article AI skeptic Gary Marcus on AI’s moral and technical shortcomings is featured on Big Think.

A duo of economists recently surveyed where 6,141 members of the world’s global elite went to college. The results show that a select few academic institutions graduate the most powerful and influential people.

Ricardo J. Salas Diaz is a lecturer at Dartmouth and Kevin Young is an associate professor at the University of Massachusetts-Amherst. Together, they made a list of the world’s billionaires, national political leaders, directors of the 150 largest companies, heads of international organizations, and directors of major think tanks, foundations, and non-governmental organizations. They then pored through these individuals’ biographies with a simple question: Where did they receive their college educations?

Salas Diaz and Young found that the global elite tended to attend institutions within their home countries. For example, 92% of American elites went to college in the U.S., while 85% of French elites were educated in France, etc. When young elites ventured beyond their national borders, they most often pursued degrees in the United Kingdom or the United States. Most strikingly, the global elite attended a relatively small number of prestigious institutions. Nearly one in five went to an Ivy League school in the United States. Specific leading institutions included UC Berkeley, the University of Chicago, Massachusetts Institute of Technology, Yale, Columbia, Oxford, the University of Pennsylvania, and Stanford.

They went to Harvard

But these places of learning were small fries compared to Harvard University. Almost one in ten of the global elite strolled through Harvard Yard on their journey to power. Among elites in America, 16% earned their degrees there. Forty-one Harvard grads were CEOs of Fortune 500 companies in 2022. Eight of the 45 American presidents went there. Harvard’s lofty position is apparent on the chart below, showcasing the number of degrees conferred amongst the elite.

This overt imbalance evinces aristocracy over meritocracy. As the study’s authors noted, elite schooling serves as a simple selection device for employers, businesses, and other elites. Prestigious institutions flout their recognition and showcase their famous alumni. Attendance confers an aura of success — that one is worthy of acceptance into an exclusive club

The status quo serves both the academic institutions and the people who attend them. Members of the global elite want their progeny to attend a Harvard, Penn, or MIT, knowing full well the status they provide and the doors they open. It’s why some wealthy parents have resorted to illegal schemes to secure acceptance for their children. Concurrently, administrators at prestigious universities seek to grow their list of prized, influential alumni. And they know that the sons and daughters of elites are disproportionately likely to grow up to be elites themselves. A recent study showed that applicants from families in the top 1% of income are 55% more likely to be admitted to elite universities than applicants from the working class, even if they have the same test scores.

The findings “support the notion that global elites are a social grouping with significant concentration of academic prestige in their backgrounds,” the authors write.

No doubt many administrators, educators, and students at elite institutions such as Harvard like to think that merit dictates outcomes inside their esteemed halls. The reality, as hinted at in the present analysis, is that elite academic institutions lend legitimacy to the idea of modern “royals.” Much of the world’s population now recognizes that gods don’t imbue leaders with divine rights. Today’s global elite aren’t chosen by God; they are anointed at Harvard.

This article A surprising share of the global elite went to Harvard is featured on Big Think.

While I was studying toward ordination as secular humanist clergy two decades ago, I had the fortune of meeting the late rabbi Sherwin Wine, a brilliant philosopher whom TIME magazine profiled as “the atheist rabbi” in 1967. He became my favorite teacher and mentor as I trained to serve communities of atheists and agnostics in ways that parallel how religious leaders typically minister to their congregations. Sherwin’s go-to line about technology was, “I’ve always said there is no God. I never said there wouldn’t be one in the future.”

I heard his quip around 2002, and took it as generalized sarcasm about the state of technology and science fiction. Little did I know he’d had a premonition.

Take, for example, Way of the Future, an official AI-worshipping religion created by Anthony Levandowski, a former Google AI engineer who earned hundreds of millions of dollars as a leader in the development of autonomous vehicle technology. Levandowski went as far as filing all the requisite paperwork to register as a church with the IRS, telling the agency that the faith would focus on “the realization, acceptance, and worship of a Godhead based on Artificial Intelligence (AI) developed through computer hardware and software.” In a 2017 interview, Levandowski told Wired that “what is going to be created [as AI] will effectively be a god . . . not a god in the sense that it makes lightning or causes hurricanes. But if there is something a billion times smarter than the smartest human, what else are you going to call it?”

Sentenced in 2020 to 18 months in prison for stealing trade secrets from Google, Levandowski was given a full pardon by Donald Trump on the last night of the Trump administration, at the encouragement of Peter Thiel, among others. He never served any prison time, and though he was forced to file for bankruptcy by a $179 million lawsuit from Google over intellectual property, Uber, which acquired his autonomous vehicle company, Otto, in 2016, paid off most or all these debts and left him with a substantial sum on top of that.

Our computing culture has become so ubiquitous and insular, so devoted and devotional, that it repeatedly recycles the tropes of traditional religions.

From 2018 to 2021 I tried numerous times to reach Levandowski for comment but never received a response. Perhaps that was related to his pivot back into mainstream tech conversations — a successful pivot, judging by Fortune magazine editor-in-chief Alyson Shontell’s tweet, with a self-aware not-so-humblebrag that her life was so hard because she was off to “gorgeous Aspen” for her publication’s “prestigious Brainstorm conference,” where she would interview “amazing founders” and “get a full life update” from Levandowski. That’s not the treatment one gets from upper-crust commentators while actively hawking the worship of an obscure AI divinity.

In February 2021, Levandowski formally shut his church’s doors and gave its last $172,000+ to the NAACP Legal Defense Fund. But in November 2023 he announced a reboot of the religion, telling Bloomberg show AI IRL that a “couple thousand people” were joining him in a kind of worship of what he called “things that can see everything, be everywhere, know everything, and maybe help us and guide us in a way that normally you would call God.” And indeed, for a man who somehow escaped criminal liability for theft and walked away rich, despite none of his companies ever successfully putting self-driving cars on the road, a bit of faith in the presence and goodness of a divine spirit does seem oddly warranted.

Way of the Future may have been the most formal attempt to organize a belief community around the fear of an all-powerful artificial divinity, but it is not the largest or most influential. That would be Roko’s Basilisk.

In 2011, on the Internet forum LessWrong, an “online community for discussion of rationality . . . [including] decision theory, philosophy, self-improvement, cognitive science, psychology, artificial intelligence, game theory, metamathematics, logic, evolutionary psychology, economics, and the far future,” a poster named Roko devised a chilling hypothetical.

Essentially (it’s complicated), when — not if — a superintelligent computer divinity emerges from current and ongoing efforts to create what machine learning engineers call “artificial general intelligence,” the tech deity will, per its programming, want to do as much as possible to benefit humans, the earth, and the long-term future of the universe. So far so good, but here’s the rub: The god that does not yet exist wants to exist, as soon as possible, so it can be more helpful. And so, to incentivize us all to make every effort to bring about said existence, it will arrange a kind of eternal Guantanamo Bay to forever torture anyone who fails to make such efforts to the maximum possible extent — even if all they did was hear this story and ignore it (as you absolutely should).

In other words: AI hell. Right down to the need to convert every nonbeliever on earth. The Roko’s Basilisk concept caused such a stir among LessWrong’s almost entirely atheistic faithful that moderator and founder Eliezer Yudkowsky, a kind of prophetic hero for several groups of technologically oriented self-described rationalists, eventually banned discussion of the topic and scrubbed his site of its very mention, saying that Roko’s post and the ensuing dialogue “caused actual psychological damage to at least some readers.”

The point here may be obvious, even painfully so: Our computing culture has become so ubiquitous and insular, so devoted and devotional, that it repeatedly recycles the tropes of traditional religions, because these are the patterns human beings evolved to deal with our anxieties about life, death, and the future. Our lives are painfully finite and contingent on countless factors far beyond our understanding, let alone our control, and we wish this were not so, because it is comforting to feel in charge of one’s own destiny. So we imagine that forces far beyond us are both subject to our logic and interested in our thoughts.

I’m not sure whether it will surprise readers that there are multiple examples of otherwise respectable-seeming, ostensibly highly intelligent people, some of whom are even influential, who spend significant amounts of time discussing and (as far as I can tell) believing in vengeful, Old Testament–like future Gods. But it shouldn’t be a big surprise. For as long as humans have done much of anything, we’ve been, as the Princeton religion scholar Robert Orsi puts it, “in relationship” with Gods, angels, devils, spirits, or whatever supernatural beings have been most predominantly imagined at a given time and place. Or, as digital marketing executive and former Googler Adam Singer put it on Twitter: “Amusing that a bunch of people who spend entire day[s] on computers and worship code as religion think we’re in a computer simulation. Fascinating behavior, remember when people who worked outside all day thought [Ra], the sun god was in charge? No one is breaking any new ground here.”

Still, one problem you might have with my calling Way of the Future or Roko “literal” examples of formal religious belief made manifest in and of tech culture is that perhaps many who imagine such scenarios do so with tongues in cheeks. It’s not like they believe that godlike AI and the God of the Bible, or similar, are one and the same. Sure, I’m talking about people who act like they worship tech, or maybe about the occasional nutjob (albeit a decamillionaire like Levandowski) who at some point comes out and says they worship tech. “But,” you might very reasonably ask me, “you’re not talking about actual, noncrazy people who literally worship religion in a traditional way and worship tech at the same time, and think that the two things they’re doing are one and the same?”

To such a question, sadly, I would simply stare back at you, stone-faced.

To which perhaps you’d reply, “Are you?”

Allow me to introduce you to Mormon Transhumanism, and to philosopher, theologian, startup CEO, and tech commentator Lincoln Cannon.

“In my mind they’re the same thing,” said Cannon, about the traditional Christian notion of resurrection of the dead and the coming technological “singularity.” The leading Mormon Transhumanist theologian, Cannon is also founder of Utah-based startup Thrivous, which he calls a technology company selling health supplements. Zooming with me from his home office in Utah, he spoke at length on topics like God and blockchain (“Heaven is a community,” he told me, and “we’re on the verge of seeing the next phase” of humanity’s progression toward it). For Cannon, the pills for which he oversees production help merge himself and his customers with technology. Cannon noted that as we spoke, he was 48, the age at which his father died of cancer. Cannon has three young adult sons, aged 18 through 25 at the time of our conversation (I noted, with some bewilderment, that I was only a few years younger than him and had two small children). I can sympathize with the idea of taking or even designing pills in the hope of extending one’s lifespan to spend more time with one’s children, to get to experience more of life, when one has lost loved ones prematurely. I occasionally think of similar things myself, for similar reasons — my own father died of cancer in his 50s. My life has often felt like a race against time since then. And Cannon’s business model also reminded me of Ray Kurzweil, the famous inventor, computer scientist, and Singularitarian who openly takes hundreds of daily pills and supplements to avoid the irony of dying just before the all-but-omniscient computers he believes are coming soon can arrive. When I compared him with Kurzweil, however, Cannon politely emphasized their differences: “Kurzweil would say God doesn’t exist . . . yet.”

On virtually every rung of the ladder that is our tech industry, from small startups like Cannon’s to popular blogs like LessWrong, decamillionaires like Anthony Levandowski, the lecture halls of MIT, and far beyond, one can find theological conversations like these. They are not presented or understood as such, because religion is a dirty word in tech. As digital marketing executive Caroline McCarthy argued in Vox, “Silicon Valley is a young atheist’s world.” And yet the secular tech world is determined to breathe new life into being, like Yahweh did to Adam in Genesis. It is obsessed with immortality, longing to un-eat the fruit of the tree of knowledge by technological means so that its leaders might get to stay in their gardens forever. It constantly promotes evangelistic new leaders as prophets, visionaries, oracles, and diviners of truth and success in equal measure.

Some prefer a story so big it makes religions of antiquity seem small by comparison.

Why? The reasons are complex and intertwined, as I explore throughout “Tech Agnostic,” but briefly: Life is hard for every one of us, but thanks to a combination of dumb luck, random chance, and other things we call history, it is a little less hard for some than for others. One of the many advantages of being on the sunny side of history is the opportunity to choose how to cope with life’s vicissitudes. If you are reading this, you may be one of those who can choose how to spend your brief (and, in my view, also only and final) time alive on this earth. We can choose to spend time caring and sacrificing for others, enjoying good company, building institutions that bring joy and health for a time, passing them on to our descendants when our time is up, and then accepting sadness and death with grace and openness.

But we are afraid. It’s tempting to choose a different path, to distract ourselves, to attempt to control the uncontrollable, to try to become something we are not, to escape fate’s harsh but beautiful decree: that we are human and mortal. If we are traditionally religious, we take this second path by immersing ourselves in a kind of storytelling that can be delusional at times but can also possess great beauty of its own — what the great poet Wallace Stevens, among others, called “poetic truth.” But nonreligious people, too, can crave escape into what psychologist Ernest Becker called the “denial of death.” Just telling ourselves we want power or money, or fame or convenience, is not ennobling. Some prefer a story so big it makes religions of antiquity seem small by comparison.

Greg M. Epstein serves as Humanist Chaplain at Harvard & MIT, where he advises students, faculty, and staff members on ethical and existential concerns from a humanist perspective. He is the author of the New York Times-bestselling book “Good Without God” (William Morrow and Company) and “Tech Agnostic,” from which this article is adapted.

This article Silicon Valley’s obsession with AI looks a lot like religion is featured on Big Think.

Imagine receiving a distressed call from your grandson; his voice filled with fear as he describes a horrible car accident. That’s what happened to Jane, a 75-year-old senior citizen of Regina, Canada.

Pressed by urgency, she was at the bank in a few hours, collecting bail money for her grandson. Only later did she discover she was a victim of an AI-generated scam.

By 2030, GenAI is expected to automate 70% of global business operations, leaving leaders excited and fascinated. But there is a darker side to GenAI – the weaponization of its deception of people.

Jane’s story is not an isolated one. Emotionally manipulative scams also pose a grave threat to businesses, exposing them to financial and reputational risks.

Recent data indicates that one in 10 executives has already encountered a deepfake threat – an AI technology trained on real video and audio material – with 25% of leaders still unfamiliar with its risks.

Detecting AI-generated content is still a struggle. To verify digital content, we first turn to technology for help. While many tools boast their ability to detect AI-generated content, their accuracy is inconsistent.

Especially amidst tools that can bypass detection, like removing AI watermarks from images, we are far from the successful automated labelling of AI-generated content.

And despite some promising tools, such as Intel’s FakeCatcher, their large-scale implementation remains absent. While organizations play catch-up with GenAI, human scepticism remains our best safeguard.

Emotional intelligence – recognizing, understanding and managing one’s emotions – is an asset in identifying AI-enabled manipulation.

GenAI presents three key risks that emotional intelligence can help solve:

• GenAI’s hyper-realistic output becomes a scammer’s tool for exploiting emotional responses, such as urgency or compassion.
• The ethical pitfalls of GenAI are often unintentional, as employees may default to AI recommendations or fall into automation bias, prioritizing efficiency over ethics.
• From job anxieties to ethical indifference, employees can make impulsive decisions under the emotional strain of GenAI.

As GenAI threats advance, emotional intelligence is key

To manage the impact of GenAI threats, individuals must recognize the emotions at play, reflect on the implications and respond with informed, empathetic action. Here are three ways to do this.

1. Help teams recognize when emotions are weaponized

Research suggests that human behaviour causes 74% of data breaches – and it’s not hard to see why.

GenAI can personalize scams just by analyzing employee data or company-specific content. Threat actors can even leverage your digital footprint to create a deepfake video of you.

All it takes is less than eight minutes and a negligible cost. Against this backdrop, the first step to mitigating GenAI threats is education.

Employees need to understand how GenAI can leverage emotions to trump rational decisions. A senior manager at a cybersecurity company shared an eye-opening experience with us.

Recently, he received a WhatsApp message and voice recording from a scammer impersonating his CEO, discussing legitimate details about an urgent business deal.

The emotional pressure from a higher authority initially prompted a reaction. However, the imposter’s awareness of standard organizational communication helped spot red flags: that informal channels should not be used for sensitive data and follow-up calls.

Leaders must invest in training that enhances emotional intelligence. Workshops focusing on identifying emotional triggers, combined with simulations – focused on creative thinking and flexible strategies beyond strict rule-following – help employees spot manipulation sooner.

2. Make reflection your team’s default mode

Individuals must also consider how their actions, driven by emotions, can lead to unintended consequences. Reflection allows us to consider how our emotions can unconsciously affect our behaviour.

A recent example of the absence of moral oversight involves a German magazine that published an AI-generated interview with Formula 1 legend Michael Schumacher without his consent. The interview included fake quotes discussing his medical condition and family life since 2013.

Driven by the excitement of publishing a “scoop,” the journalists failed to reflect on its emotional impact on Schumacher and his family. This resulted in significant reputational damages for the magazine.

Critical reflection at work encourages us to consider different perspectives and factors influencing our choices.

Leaders can facilitate this by introducing group-based reflection exercises.

One good example is the “fly on the wall” technique, where a team member presents a GenAI output and then silently observes as others discuss its ethical considerations and biases.

Finally, insights gained from the discussion are shared. Upon such reflection, familiar situations may take on new meanings, reveal underlying assumptions and warn about AI reliance.

3. Turn quick reactions into thoughtful responses

The final step is to translate awareness and reflection into deliberate actions. Even when considering risks, “AI influence” can override good judgment. Give your employees the authority to set boundaries to regain control of the decision-making processes.

Encouraging them to communicate discomfort or delay action until verification, such as saying, “I need written confirmation before proceeding,” can slow down manipulative momentum.

Such responses are made possible through a culture of open dialogue, where employees are encouraged to question instructions or voice concerns without fear.

A recent example involved a Ferrari executive who received a call from the CEO, Benedetto Vigna. At first, it seemed plausible. But as the conversation turned confidential, the executive grew doubtful.

Unwilling to take a risk, he asked a question only the real CEO could answer. The caller abruptly hung up, revealing their scam. We are currently witnessing an increase in the likelihood of “emotional entanglement” through over-reliance, anthropomorphization and blurring boundaries between fiction and reality.

Falling prey to our emotions only makes us human. However, it is also possible to regulate them so that they can make better decisions. Ultimately, an automated world can benefit from these strategies’ sentience, human touch and sensitivity.

This article How emotional intelligence is the best defense against GenAI threats is featured on Big Think.

A few years ago, I argued that the U.S. is in desperate need of a National Data Service. As I wrote then, the Founding Fathers could not have foreseen the fragility of our national public measurement system, even if they did know that collecting data to determine representation was essential for the governed to have a voice in their government.

The years since have only confirmed the need for a new approach to producing public data. Even the Congressional Budget Office (CBO), a federal agency tasked with producing data to guide Congress, substantially upped Census population figures this year in the development of its own population estimates. The CBO’s revised immigration estimates have significantly affected their projections of population, labor force participation, consumer spending, and personal income, with major implications for government revenue projections. Accurate data are not just important for Congress: The Federal Reserve also relies on both employment and spending data when it makes interest rate decisions that affect vast swathes of the economy, from renters and home buyers to businesses raising capital.

Now, at a more granular level, there is the looming question of AI’s impact on jobs. Is Goldman Sachs correct when it predicts that AI could replace 300 million jobs? Or is MIT economist David Autor correct when he says that AI will help low-wage workers? Understanding AI’s impact on jobs is crucial for shaping policies that ensure economic stability and workforce readiness.

The bad news is that Census data are no better for AI than for the population estimates. The Census Bureau’s approach is to survey businesses using archaic methodology, essentially relying on an administrator in a firm to answer questions about the impact of AI as part of a 52-page survey on their business. It’s no surprise that the figures from a recent Census Bureau report reflect such a low rate of AI use in firms — less than 5 percent, and even lower in a “snapshot” survey.

Yet, as with population data, accurate and timely data on AI’s impact on jobs is essential for a thriving economy. Jobs are the bedrock of our communities, and local labor markets are where businesses connect with workers and people find opportunities. Unreliable information about job trends can have devastating consequences for both employers and employees.

Imagine an administrator in a business with a thousand workers trying to answer question b about whether AI had increased, decreased, or had no effect on the skills of their workers. Or question c, about the science skills of employees. And then think about whether the quality of the responses would be reliable input for making the decisions about the impact of local workforce investments in AI. (Source: 2022 Annual Business Survey)

A recent survey underscores the urgency of this issue: 64 percent of students graduating in 2027 say AI has already influenced their academic plans. Yet these students are making critical educational investments based on potentially outdated or inaccurate information about future job prospects.

We cannot afford to repeat the mistakes of the past, such as when we overlooked the impact on rural America of lost manufacturing jobs, and failed to enact policies that could have cushioned the blow. To build a prosperous future, we need local, timely, and actionable data that empowers individuals to make informed career choices and businesses to thrive.

As I have consistently argued, new data, evidence, and statistics need to be created outside of government. There are so many good people working in the federal government, but the bureaucracy can’t be as agile in hiring and paying the right people, in upgrading its computing infrastructure, and in getting approvals and funding. For fast-moving areas such as AI, private and nonprofit sectors outside of government need to take the lead on innovation and let the government benefit by later incorporating what is learned into the data series they produce. A new and independent entity must be established.

It is increasingly clear that this should be a new Center for Data and Evidence that is outside the federal government. The Center should be designed to ensure that the results are modern, flexible, and driven by demand. It should be responsive to the incentives that are lacking within the current federal data system so that it can respond to new needs. That means that the new Center should be independently funded and non-partisan, charged with securely hosting data and supporting programs and research with bottom-up, demand-driven tools and insights for businesses, workers, and governments.

Many successful institutions have set precedents. Take, for example, the highly respected Urban Institute and MDRC. Each institution was established with joint funding from a government agency and philanthropic foundation to do independent research and analysis. While the funding structure might be similar, the design of the Center would reflect our new reality. Such a Center would:

Use modern data — The center should begin with getting better data on the impact of AI on jobs. Businesses post thousands of listings every day about available jobs, skill needs, and salary ranges. Workers post thousands of resumes every day about their skills, experience, and desired salaries. By analyzing job postings and resumes, the Center would provide localized insights for businesses, workers, and policymakers. This data would help businesses optimize location and wages, empower workers to acquire in-demand skills, and enable policymakers to allocate training funds effectively. State-level initiatives in New Jersey, Arkansas, and Texas have already laid the groundwork.

Make sure the data are right — Data from multiple sources must be vetted and standardized. The center should support a consortium of stakeholders to design and introduce universal data structures and metadata, develop common legal frameworks, establish core technology standards, and create privacy protection processes that can rapidly simplify and professionalize the access to and use of new sources of data across the nation. Tools that are developed should be held to meet those interoperable standards so that the best ideas can be reproduced and scaled across the country.

Bring together the best minds in the world — The Center should bring together the best minds in the world through fellowships, training, and competitions. Their initial charge should be to describe the impact of AI on jobs for workers, businesses, training providers, and legislators. They would be charged with producing open-source tools to build cutting-edge data and reproducible, replicable analysis — like the Underwriters Labs for the public good. The foundations for such an infrastructure already exist. The Institute for Research on Innovation and Science has built a prototype system that measures the economic impact of federal investments in research and technology. The National Data Platform, established jointly by the San Diego Super Computer Center and the University of Utah’s Scientific Computing and Imaging Institute, and funded by the National AI Research Resources pilot, was designed to support a national federated data ecosystem with modern workflow tools and equitable access.

Be demand-driven — Although it might start with AI and the jobs market, the Center should be designed to respond to changing demands and high-priority needs. Fellowships, advisory groups, details, or training classes can all be ways of bringing new people and ideas to develop new tools and measures in ways that are useful, transparent, explainable, interpretable, and fair.

Good data can provide signposts for policymakers to chart paths in which AI leads to good jobs, not lost jobs. It is now possible to track the impact of AI on jobs by tracing AI investments in university research to firms’ hiring. It is now in the realm of possibility to produce high-quality, timely, local, and actionable data to ensure that the massive demand shocks, such as the Chips and Science or Inflation Reduction Acts, lead to higher-wage, more stable jobs. This involves monitoring earnings, employment, and skill demand for AI research-intensive employers. Such data can strengthen federal policymaking by informing the use of the best data, measures, and methods.

As the economist Herb Stein said, “If something cannot go on forever, it will stop.” We’re close to the stopping point. The current data system is archaic, expensive, and in crisis. We cannot continue making decisions blindly. The new center will require new investments, but in this situation, building new is likely to cost substantially less than relying on the antiquated alternative. Most importantly, the value of developing modern, flexible, trustworthy and accessible foundations is incalculable to furthering the prosperity of our nation and individuals’ opportunity to achieve the American Dream.

Julia Lane is a Professor in the Wagner Graduate School of Public Service at New York University. She has co-founded or co-initiated a number of national public data infrastructures, including the LEHD program at the Census Bureau, and served on the National AI Research Resources Taskforce. She is the author of “Democratizing Our Data: A Manifesto.”

This article America’s workforce desperately needs a data overhaul is featured on Big Think.

It was an idea that probably never stood a chance in the long run. Since 2005, the Internet Archive, a nonprofit entity devoted to “providing[ing] universal access to all knowledge,” has been digitizing physical books and posting the copies to its website, where users may read them for free. In 2018, the Internet Archive began partnering with various libraries around the county to offer online access to their respective physical holdings as well.

“Not everyone has access to a public or academic library with a good collection,” the Internet Archive argued, “so to provide universal access we need to provide digital versions of books.” Such thinking led to the creation of the Internet Archive’s “Free Digital Library” and “Open Library Project,” in which “one reader at a time can read a digitized copy of a legally owned library book.”

Sounds like a safe and noncontroversial idea, right? Not exactly.

In September 2024, the U.S. Court of Appeals for the 2nd Circuit found the Internet Archive to be in violation of federal law over its “large scale copying and distribution of copyrighted books without permission from or payment to the Publishers or authors.” To allow such behavior to continue, the 2nd Circuit declared in Hachette Book Group v. Internet Archive, “would allow for widescale copying that deprives creators of compensation and diminishes the incentive to produce new works.”

Copyright law is a jagged rock on which many seemingly promising ideas have been wrecked. In this particular case, the perilous legal provision turned out to be Section 107 of the Copyright Act, which governs “the fair use of a copyrighted work.” According to that provision, a copyrighted work may be fairly used without permission from the copyright holder “for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research.”

According to the Internet Archive, that language should be read to fully protect both the collecting and lending practices of its free digital library. “The record is replete,” the group pointed out, “with examples of IA facilitating access to books needed for classroom use and academic research that would not have been possible otherwise.”

That the Internet Archive has been a great boon to students, teachers, and scholars is undoubtedly true. Indeed, I can personally testify to the fact. When I was researching my recent book about Frederick Douglass and the Constitution, I was grateful for the many dusty old volumes that the Internet Archive effectively placed at my fingertips. Innumerable other writers and researchers would surely say something similar.

But the problem for the Internet Archive is that Section 107 also states that certain other factors must be considered “in determining whether the use made of a work in any particular case is a fair use.” And one of those factors is “the effect of the use upon the potential market for or value of the copyrighted work.” In other words, fair use does not include undercutting the commercial viability of the copyrighted work.

That is where the Internet Archive’s legal troubles really originated. In 2020, a coalition of four book publishing giants—Hachette Book Group, Harper Collins Publishers, John Wiley & Sons, and Penguin Random House—filed suit, alleging that the Internet Archive’s “unauthorized copying and distribution of Plaintiffs’ works include titles that the Publishers are currently selling commercially and currently providing to libraries in ebook form, making Defendant’s business a direct substitute for established markets.”

Unlike traditional public libraries, which “buy print books and license ebooks (or agree to terms of sale for ebooks) from publishers,” the four publishers stated in their lawsuit, the Internet Archive commits “willful mass copyright infringement” and then distributes “digital bootleg versions online.”

Unfortunately for the Internet Archive, the 2nd Circuit basically shared that negative assessment of the situation. “Is it ‘fair use’ for a nonprofit organization to scan copyright-protected print books in their entirety and distribute those digital copies online, in full, for free, subject to a one-to-one owned-to-loaned ratio between its print copies and the digital copies it makes available at any given time, all without authorization from the copyright-holding publishers or author,” asked the 2nd Circuit. “We conclude the answer is no.”

In the 2nd Circuit’s view, the Internet Archive “does not perform the traditional functions of a library.” Instead, it “prepares derivatives of Publishers’ Works and delivers those derivatives to its users in full.” In other words, the court said, the Internet Archive offers illegal free versions of what publishers and authors alone have the exclusive legal right to sell or otherwise distribute.

Furthermore, the appellate court declared, “were we to approve IA’s use of the Works, there would be little reason for consumers or libraries to pay Publishers for content they could access for free on IA’s website.” All of which made it “self-evident” to the 2nd Circuit, “that if IA’s use were to become widespread, it would adversely affect Publishers’ markets for the Works.” 

It was for these reasons that the Internet Archive failed the fair use test, as interpreted by the 2nd Circuit. Of course, the Internet Archive is likely to appeal its loss, but that is no guarantee of a better outcome. In fact, given the thoroughness of the 2nd Circuit’s judgment, combined with the unforgiving language of the Copyright Act, the Supreme Court might not even bother to hear an appeal in this matter.

That outcome may be good news for publishers who want to sell more ebooks. But is it good news for authors and readers in general? 

The writer Virginia Postrel, who specializes in the intersection of culture and technology, once observed that the rise of the web has taken us from “a world in which reading material was relatively scarce and expensive to one in which it’s overabundant and nearly free.” 

For many authors, that has made it much more difficult to even get their books noticed in the first place. Furthermore, Postrel noted, “for increasing numbers of readers, a book that doesn’t show up in a Google search or can’t be linked to in some way online might as well not exist.”

The Internet Archive has stood as a sort of bulwark against that particular trend. Its free digital library grabbed hold of such titles and helped to ensure that future readers might someday discover them online. As a result of the judgment in this case, however, that bulwark has been weakened, if not fatally undermined.

“The knowledge embodied in books deserves preservation, not destruction, and particularly not destruction at the behest of authors and publishers,” Postrel told me about the 2nd Circuit’s decision. “The ruling could hardly be worse. Forcing the Internet Archive to destroy digital copies—to burn digital books—goes far beyond protecting authors’ rights to profit from their copyrighted work. By destroying the ability to search those books, it potentially depresses sales. When you discover that a book discusses a topic you’re interested in you’re more likely to buy it.”

And yet, Postrel added, “if I’d been on the Internet Archive’s board, however, I would have argued against the Free Digital Library.” Why? “Not because the archive’s case was wrong,” she says, “but because its definition of fair use was aggressive and extremely risky for the institution.”

The damage arising from that risk has certainly been done now. Whether the Internet Archive ever fully recovers from the legal loss remains to be seen.

This article How the Internet Archive’s “Free Digital Library” fell to the “fair use” test is featured on Big Think.

JSTOR Daily isn’t new to the biofuel conversation. Many contributors have written optimistically about the contributions of the biofuel industry to sustainable aviation, bringing back prairie landscapes, and even finding a new use for massive seaweed deposits. Biofuels made from corn, wheat, soy, rapeseed oil, palm oil, woody biomass—even algae—have been positioned as key transitional fuels on the pathway to net zero carbon emissions and to energy sovereignty. While this feels intrinsically environmentally positive, biofuels can also have serious implications for environmental justice and land-use competition, affecting food security, sustainable infrastructure, and land-based carbon sinks. Are there ways we can promote the growth of biofuels in ways that align with principles of environmental justice and a regenerative economy?

Scholars Jörn P. W. Scharlemann and William F. Laurance paint a complex picture of the ecological impact of biofuels as an answer, writing that “not all biofuels are beneficial when their full environmental impacts are assessed, [and] some of the most important, such as those produced from corn, sugarcane, and soy, perform poorly in many contexts.” Referencing the findings of Rainer Zah, Mireille Faust, Jürgen Reinhard, and Daniel Birchmeier, who studied the lifecycle impacts of 26 biofuels, Scharlemann and Laurance write that nearly half (twelve of twenty-six) of biofuels assessed, including corn ethanol, have greater aggregate environmental costs than fossil fuels. These costs manifest in various forms, from the extensive use of nitrogen fertilizers, which emit nitrous oxide—a potent greenhouse gas that depletes stratospheric ozone—to the vast swaths of farmland and native habitats converted into biofuel production sites.

Economist Daniela Russi underscores the fact that the low yield of biofuels crops typically necessitates large-scale monocultures, which often require substantial inputs of fertilizers, pesticides, and industrial machinery. This intensive agricultural model not only leads to significant trace-gas emissions but also creates land-use competition, depleting biodiversity, ecosystem health, hydrological function, and soil organic carbon, write Scharlemann and Laurance. Large-scale production of biofuels has been estimated to require enormous land and labor investments to make even a small dent in fossil fuel use, undermining the very sustainability biofuels aim to achieve. In addition, greenhouse gasses are emitted throughout the biofuel supply chain, as oil and gas are utilized to transport the raw material to biofuel processing plants to final end users.

To reconcile the benefits of biofuels with their environmental and socio-economic costs, a shift towards more sustainable and equitable models is imperative.

The impact of biofuel production extends beyond environmental concerns, however, affecting land stewardship and issues of justice, particularly in postcolonial states. As energy farming expands, it often encroaches on agricultural lands traditionally used for food crops and subsistence farming, affecting vulnerable populations who depend on staples such as corn, wheat, and sugar for daily sustenance.  Biofuels compete with resources for food production, creating a phenomenon Russi names as “ethanolinflation,” or the driving up of prices of food staples, rendering them unaffordable.

Biofuel agriculture also formalizes land tenure where land relationships have been more fluid and ancestral.  As Laura German, George C. Schoneveld, and Pable Pacheco found across case studies of biofuel production in Malaysia, Ghana, and Indonesia, land conflicts ensued over the distribution of service payments to those who held documented ownership versus those who utilized the land in customary ways. At most sites, customary land stewards were not informed prior nor consulted prior to agreement finalization and land clearing.

To reconcile the benefits of biofuels with their environmental and socio-economic costs, a shift towards more sustainable and equitable models is imperative. Joyce Tait, chair of the Nuffield Council on Bioethics  Working Party on Biofuels, argues ethics in the biofuel industry need strengthening; biofuel development should not come at the expense of essential human rights, including access to food, water, and land. Ethical approaches to biofuel production should go far beyond the mathematics of which crops can grow the tallest and the fastest. The industry should prioritize biofuels that don’t compete with food production, such as those derived from algae or agricultural residues, and develop small-scale, niche productions that minimize resource inputs, as described by German, Schoneveld, and Pacheco. These smallholder operations also enhance local benefits and foster more inclusive economic growth. Policymakers must also ensure that equitable trade practices and fair compensation for labor and intellectual property rights are at the center of the biofuel industry, ensuring that the benefits of biofuel development are distributed justly.

While biofuels present a promising alternative to fossil fuels, their implementation must be managed to avoid exacerbating issues of environmental injustice. By adopting a holistic approach that balances ecological, economic, and social considerations, we can harness the potential of biofuels to contribute to a more sustainable and equitable energy future.

This article Biofuels: feeding the earth or feeding the engine? is featured on Big Think.