I’m the kind of person who shares a genuine compliment often, telling someone who is standing behind me in the airport line that I love their sneakers or letting one of my daughter’s friends know they have a beautiful smile. I also like to do nice things for people, like bringing the veterans that I work with doughnuts and coffee every Monday morning.

I learned to be kind from my grandfather who raised me. He drove the nuns from my Catholic grammar school to their doctors’ appointments, and he shared smiles and jokes with grocery store clerks, neighborhood kids, and almost every soul whose path he crossed. I wanted to follow in his footsteps and engage as often as I could in my own random acts of kindness. I recognized that being kind and giving selflessly to others made me feel good as well, but it wasn’t until recently, 25 years into my career as a trauma psychologist, that I learned about the science behind random acts of kindness.

Research indicates that engaging in random acts of kindness boosts happiness in both those who give and those who receive. A systematic review of 24 studies indicated that performing acts of kindness boosts the well-being of the giver.

One study on the effects of a seven-day intervention—in which people intentionally tried to do nice things for others—found that it didn’t matter if someone was giving to those with whom they had strong ties (family members or friends), or those with whom they had weak ties (strangers). All had equally positive effects on happiness.

Oliver Scott Curry is the research director for Kindlab, a team of researchers who study the causes and consequences of kindness. The group offers tools and programs to help increase kindness; you can even complete the Kindness Quotient, a research-based questionnaire that takes less than five minutes to fill out and gives you an indication of how kind you are compared to others.

I contacted Curry and asked him what he thought about the value and opportunities for kindness in what feels like an incredibly hostile world. “Humans are naturally kind, under the right circumstances,” he tells me.

Indeed, according to research across diverse cultural and economic settings, from the U.S. and Canada to Uganda, India, and South Africa, the well-being rewards one experiences from helping others appears deep-rooted in human nature. 

Jonathan Passmore, a psychologist and coach in the U.K., has developed a positive psychology coaching technique called CAKE, which stands for consistent acts of kindness and empathy. I asked him how engaging in CAKE can help teach someone how to spread kindness. “The list of possible actions is limitless,” he says. “I advocate for folks to initially commit to a one-week plan—choosing to engage in one act of kindness per day.” On Monday, for example, you can greet the barista at your local coffee shop by name, and wish them a pleasant day, he offers. On Tuesday, “you could write a handwritten card to a colleague and thank them for something they have done.” Do this for a week, and “one often gets hooked and begins to hold an empathic stance towards most they meet.”

You’re never too young to start. Kindlab offers kindness challenges—skill-building, games, and more—for younger family members. One of my favorites is the kindness walk: a map that reminds and encourages kids to smile at someone they don’t know, give someone a sincere compliment, or pick up litter the next time they take a walk. 

Givers often underestimate the positive effect such gestures have on receivers. Undervaluing our kindness may discourage us from engaging in such acts. But engaging in acts of kindness is a straightforward, low-cost, widely available, and valuable tool we can all use with our family, friends, community members, and strangers. 

As a psychologist who cares very deeply about people’s well-being, I see engaging in acts of kindness as a win-win. Whether it’s buying a cup of coffee for the person in line behind you at the drive-thru, holding the grocery store door open for someone whose hands are full of heavy bags, or saying hello to a stranger, you will likely boost your own and others’ emotional well-being.

Some relationships need to end. 

Years ago, I treated a woman whose husband was deliberately trying to drive her toward suicide while attempting to extract money from her family. What struck me was not only the cruelty of his behavior, but his complete lack of empathy or remorse for the harm he was causing. Leaving him was necessary, and the woman is now happily remarried.

Most troubled relationships look nothing like that. They involve disappointment, resentment, misunderstanding, conflicting needs, old wounds, and people who have never learned how to reconcile after hurting one another. In more than 20 years of clinical practice, that extreme case remains the only time I directly recommended divorce.

Mental-health professionals have become remarkably good at helping clients identify unhealthy relationships. We have a sophisticated vocabulary for boundaries, trauma, toxicity, narcissism, gaslighting, and emotional safety. But I worry that we are far less effective at helping people navigate difficult but meaningful connections.

There is an irony here. Psychotherapy itself is predicated upon the healing power of human connection, yet therapists can be surprisingly quick to undervalue relationships outside the therapy room. Furthermore, relationships themselves teach basic psychological capacities such as regulating emotions, tolerating disappointment, and negotiating differences—skills which therapy is supposed to cultivate.

Too often, we mistake relational distress for relational dysfunction. Wall Street Journal columnist Elizabeth Bernstein recently highlighted this eloquently. She reported on therapists who encouraged adult children to go “no contact” with their parents, often after surprisingly little effort to mend ruptures. In a survey of roughly 7,000 estranged parents, one-third believed their adult child’s therapist had influenced or even recommended the decision to cut ties. That should give us pause.

The issue extends well beyond parents and children. Similar patterns appear in marriages and among siblings, friendships, and other close bonds. Therapists have become too quick to support ending difficult relationships instead of helping people develop the skills to repair them.

In reality, walking away from another person often solves one problem while creating others. Increasingly, my practice sees clients who have ended relationships that took a lifetime to build. The immediate conflict may be gone, but so is the connection. Many become lonelier and less equipped to navigate conflict, disappointment, and repair elsewhere.

Tragically, many who end contact with loved ones also become more distressed in the long run, since close relationships are a foundation of mental health. Maintaining social connection is a form of “fitness” that requires ongoing cultivation. Even when strained or difficult, human bonds provide belonging, support, identity, and resilience. That may matter more now than ever, as many meaningful in-person social interactions are dwindling in the face of digital alternatives.

I learned this lesson the hard way about 15 years ago.

A woman came in because she was deeply concerned about her husband, who had obsessive-compulsive disorder. We eventually got him into treatment. He worked hard, completed difficult exposure therapy exercises, and improved considerably. By the standards I had been trained to use, the treatment was successful. But their marriage was not. His wife remained deeply unhappy, and their connection deteriorated. Eventually he lost most of his hard-fought progress, and she fell into clinical depression. Looking back, I can see the mistake clearly: we treated the patient and his symptoms, but not the relationship itself.

This reflects something larger about how mental-health care is generally structured. Our system is overwhelmingly individualistic. Diagnostics assign disorders to individuals, and insurance reimbursement follows an identified patient and his or her diagnosis. Even family psychotherapy is typically covered only when it is being used to treat an individual’s mental health. That model may make sense in other areas of medicine, such as orthopedics where a fracture occurs within one body. Mental health is fundamentally different. A great deal of flourishing and suffering happens between people, not just within them, because human beings are relational beings.

A second structural problem is that mental healthcare increasingly pathologizes ordinary negative emotions. That can backfire. Excessive focus on distress tends to amplify it. As an anxiety specialist, I know this pattern well. Avoidance may bring immediate relief, but over time it reinforces fear and reduces our capacity to tolerate discomfort. 

Effective treatment does the opposite: it helps people face distress so they become more capable to withstand adversity, not less. Applied to relationships, this becomes especially critical because even the healthiest connections generate some strain. Loving spouses disappoint us, supportive parents fail us, close friends misunderstand us, and even terrific children create enormous stress. If stress itself is taken as evidence that a relationship is unhealthy, the normal difficulties of human interaction begin to look pathological.

Therefore, therapy should do more than help people identify what’s wrong with their relationships. It should build our capacity to sustain connection and repair ruptures: to tolerate disappointment, take responsibility for our own contribution to tensions, set limits without reflexively severing ties, and accept what another person may never be able to give us.

These skills matter beyond any single relationship. Learning how to reconnect with a spouse can make us better caregivers. Learning how to accept a parent’s limitations can change how we relate to our friends, work colleagues, and children. Working through difficult interactions can both preserve meaningful connections and strengthen our ability to navigate other ones.

Mel Robbins captured this idea in accessible language through her book, The Let Them Theory. Her core insight is simple: stop trying to control other people. Let them disappoint you, disagree with you, and make choices you don’t like. Then decide how you want to respond.

That message matters because a great deal of interpersonal conflict comes from trying to force others to change. We want parents to be more emotionally available, spouses to react differently, siblings to finally understand us, and adult children to make choices we approve of. Sometimes change is possible and can be requested directly upon the foundation of a loving connection. But sometimes the wiser move is to stop demanding that another person change as a condition of staying in contact with them.

Unfortunately, some of my academic and clinical colleagues have dismissed this approach. I think they are missing something critical. Leaving is sometimes necessary, but far more often the harder and healthier work is learning how to stay connected to imperfect people without surrendering ourselves, or trying to control them.

Human beings need relationships. Relationships require skills. Therapy should help us build them.

Six months into the war between the United States and Iran, President Donald Trump has turned to economic warfare to force the country into submission and threatened an “Economic D-Day” against Iran. This new American strategy was spelled out by Treasury Secretary Scott Bessent earlier this week: a campaign of U.S. sanctions against countries and financial institutions that do business with Iran or help it evade sanctions. 

The proposed new sanctions go beyond oil exports to target digital assets, gold, and the aviation sectors that help finance the Iranian economy. Bessent described it as a campaign to “collapse every last option for Iran.” Making all this work depends on whether the U.S. can convince Iran’s most important trade partners to cut out the Islamic Republic. China, which buys around 80% of Iran’s seaborne crude oil, has already pushed back, warning that it will “take all necessary measures” to protect its interests should it be targeted by U.S. sanctions. Beijing's response casts doubt on whether Washington’s new gambit can deliver results.

If the aim, as stated, is to deal an economic blow to Iran, the fact is that the country is already under severe strain. A U.S. naval blockade has constrained oil exports, the primary source of revenue for Iran. Ordinary Iranians are facing greater economic pressure as their purchasing power is eroding: food inflation reached 128% year-on-year in July and families are cutting back on meat and other items. The dollar value of the minimum wage has fallen from about $105 to $86 since late March as the rial has weakened significantly. On Monday, Iranian currency fell to a record low of 2.02 million rials to the dollar.

So what is the goal of Bessent’s announcement? Simply put, acknowledging the reality—even if the administration won’t state it in these terms—of avoiding a further, major military campaign as the conflict drags on. For their part, Iran’s leaders have dismissed the threat of intensified economic warfare, insisting the new punitive economic measures would fail. “It is the same movie they keep playing over and over again,” remarked Iran’s foreign minister, Abbas Araghchi.

Does this mean nothing will change with the American announcement? Put another way, are we likely to see a continuation of the current state of neither war nor peace, with Iran prolonging things until it can secure a favorable deal to end the conflict? Tehran, it is true, has powerful reasons to avoid another major military escalation. A new round of fighting could deepen the damage to Iran’s energy, transport and industrial infrastructure—much of it still under repair from earlier hostilities— and would also complicate reconstruction by further straining state finances, disrupting imports and exposing an already weakened economy to another cycle of destruction.

Yet American economic saber-rattling could alter this calculus. After all, on Aug. 10, Iran’s Supreme Leader Mojtaba Khamenei appointed Brigadier General Ahmad Vahidi, a hardliner with significant experience in intelligence and operational positions, as the commander of the Islamic Revolutionary Guard Corps (IRGC), instructing him to strengthen capabilities for “powerful offensive operations against the enemy.” That language was echoed on Aug. 22 by Mohsen Rezaei, who was recently appointed secretary of the Supreme National Security Council, the top national security body of Iran. In an interview Rezaei spoke of Tehran going to introduce “changes in the conduct of the war” and bring “new capabilities” into play. He also warned that Iran had so far targeted only U.S. military bases but could begin striking American economic interests if Washington intensified its economic campaign. 

Tick tock, tick tock

If Iran is pondering a possible escalation in the military conflict, the Trump Administration appears to be betting that time may yield what bombs have so far failed to. The naval blockade in the Strait of Hormuz  remains in place, secondary sanctions are being tightened, and pressure is being increased on countries and financial networks that have allowed Tehran to circumvent previous restrictions. On Aug. 19, the United Arab Emirates, which has been a traditional financial gateway for Iran, suspended financial and commercial transactions with the country, a decision that is bound to hurt the Iranian economy.

Iran has declared the Strait of Hormuz closed, but oil has continued to move through the waterway, which is 29 nautical miles wide at its narrowest point. The U.S. military has established a protected route along the southern side of the strait, close to the shore of Oman. The Gulf states have experimented with ship-to-ship transfers and other means of moving crude around Iranian restrictions. Saudi Arabia and the United Arab Emirates have been working intensively to reduce their dependence on the strait, building pipelines and investing in storage facilities to transport their oil by other routes.

Precise estimates of oil passing through the strait are not available, but the volume seems to be somewhere between a quarter and half of the prewar flow of 20 million barrels a day. Oil prices remain elevated, hovering around $90 a barrel, but they are still far below the levels Tehran might have expected from an effective and sustained closure. Tehran had hoped that continued disruption in Hormuz would eventually make the economic costs of the confrontation intolerable for Washington and force Trump back toward compromise. Instead, the U.S. increasingly appears to believe that it can keep enough oil moving while allowing the blockade and sanctions to steadily increase the pressure on Iran.

For Iran, that creates a very different calculation. The longer the current situation continues, the greater the strain on an economy already struggling with shortages, inflation and restrictions on trade. Iranian officials are increasingly speaking of the domestic consequences. First Vice President Mohammad Reza Aref has spoken of the danger of economic discontent turning into social instability; the speaker of Parliament, Mohammad Bagher Ghalibaf, has warned that measures such as raising gasoline prices could provide an opening for renewed unrest. President Masoud Pezeshkian has acknowledged that the blockade has restricted the country’s ability to import essential goods, including fuel.

Thus, Tehran is finding it increasingly harder to sustain the condition of “no war, no peace.”

A renewed war would impose enormous costs on Iran, but its leaders might ask a different question: will those costs be greater after months of economic pressure, declining revenues and continued military containment?  If the answer is yes, then waiting ceases to be a strategy of restraint and begins to look like gradual erosion of power and strength. And they could consider escalation as a means of breaking the deadlock before Iran’s position deteriorates further.

From retaliation to initiative

Economic pressure is not the only reason Iran may be more willing to take risks. Tehran’s own understanding of deterrence has also changed over the course of the war. For years, Iranian leaders portrayed military action as retaliatory: Iran would absorb an attack, then impose costs on the aggressor. The experience of the past two years has increasingly called that logic into question. From Tehran’s perspective, repeated attempts to contain escalation did not prevent Israel and the United States from expanding their attacks.

The strategic reassessment is now visible in Iran’s military leadership and practice. After appointing Vahidi as the commander of the IRGC, Khamenei instructed him to pursue “maximum deterrence” while preparing the force for “powerful offensive operations.” The wording is significant. Offensive action is increasingly being conceived as a path to restoring deterrence. Iran has already offered a glimpse of what this could mean in practice. On Jul. 28, the IRGC launched ballistic missiles at a U.S. military base in Jordan, at a time when Trump had paused American strikes to give diplomacy more time.

In a departure from the past, the strike was not a response to an American attack. U.S. Central Command described it as an attempted surprise attack, and Washington subsequently resumed its strikes against Iran. The sequence suggested that Tehran was becoming more willing to initiate military action when it believed doing so could alter the strategic equation, rather than waiting to retaliate. Iranian leaders may also feel they have more room to act than Washington or Israel expected. Despite the enormous damage inflicted on Iran’s military-industrial infrastructure, recent satellite imagery and Israeli assessments indicate that Tehran has been restoring missile bases, production facilities, ports and other military infrastructure at a surprisingly rapid pace.

Iran has not recovered from the war, and its leaders have not become indifferent to the risks of another military confrontation. Yet capability and willingness do interact. The more Tehran believes it can regenerate the military tools damaged during the war, the less compelling a strategy of absorbing economic pressure becomes. The emerging emphasis on proactive military action and sense of military regeneration could establish a lower threshold for Iranian military escalation than Washington assumes.

If Iran considers the current balance as unsustainable, its escalation would have to differ from what it has already tried. Tehran is already attacking commercial shipping, contesting passage through Hormuz and relying on the Houthis to keep pressure on maritime traffic in the Red Sea. Intensifying those actions may raise costs, but it is unlikely to change Washington’s calculation of moving significant volumes of oil through the southern part of the strait.

Iran could resume attacks on U.S. bases, radar systems, and other facilities that support maritime surveillance and command-and-control systems that help protect and coordinate shipping through the Strait. More  dangerous still, it could attempt to strike U.S. naval vessels directly. Iran is believed to possess anti-ship missiles that, at least on paper, can threaten American warships operating in and around the Persian Gulf. The harder problem is locating and tracking moving targets with sufficient precision, why is why better intelligence, including potentially from partners such as Russia or China, could become an important variable.

So far, Tehran has generally had strong reasons to avoid inflicting large numbers of American casualties, which could trigger a much larger U.S. campaign against Iran’s military and civilian infrastructure. But that restraint depends on Iranian leaders continuing to believe that avoiding such a confrontation leaves them better off. If they conclude instead that prolonged blockade and economic pressure are steadily worsening their position, the threshold for taking that risk could fall. Tehran has already warned that major U.S. attacks on Iranian energy or civilian infrastructure would be met with strikes against comparable infrastructure across the region, potentially widening the confrontation well beyond Hormuz.

This is also why diplomacy remains the least costly way out for both sides. Washington may believe that time and economic pressure are improving its bargaining position; Tehran may increasingly believe that the same process is closing its window for action. That is a particularly dangerous combination. 

The memorandum of understanding reached earlier in the conflict is imperfect and each side accuses the other of violating it, but it still provides an existing framework for sequencing de-escalation, maritime arrangements, sanctions relief, and further negotiations. Reviving it would require concessions from both Washington and Tehran. The alternative is to allow each side’s strategy to reinforce the other’s worst assumptions until escalation begins to look less like a choice than an inevitability.

As President Donald Trump’s approval rating hit a new low, he turned once again to the art of redrawing the world map. With the stroke of a pen, the President sought to erase over 400 years of cartographic history by renaming Lake Ontario as “Lake America” through executive order. Trump’s unilateral action requires that the U.S. Board on Geographic Names update its databases to “reflect the renaming of the Lake and remove all references to Lake Ontario.”

Trump’s executive order applies to all U.S. federal maps, documents, and other communications. However, experts suggest that Trump can’t force other countries, international organizations, private mapping companies, or even U.S. state governments to use the new name. Canada’s Prime Minister Mark Carney has already responded in no uncertain terms that Canada will retain the original name of Lake Ontario “today and forever.” New York Governor Kathy Hochul also rejected the use of the name “Lake America” on New York state maps.

As two geographers who have spent decades studying political toponymy, or the politics of place naming, in both the United States and Canada, we are dismayed by the careless nature of Trump’s impulsive use of place renaming as a form of political rage baiting. 

Renaming Lake Ontario is part of a larger pattern of using place naming to promote the Trump brand and his “America First” ideology. From the very outset of his second presidential term, Trump launched the so-called “Golden Age of America” by announcing that the Gulf of Mexico would be renamed the “Gulf of America.” This has now been followed with “Lake America,” and Trump even mused about possibly renaming the Atlantic or Pacific Ocean, we presume to something along the lines of “American Ocean” or “Ocean of America.”

Over the past year, Trump’s obsession with naming places has given rise to a new term—toponymic narcissism—which refers to the condition of being obsessed with naming places after oneself.  

In many respects, Trump tends to approach the world map much like the Greek mythical figure of Narcissus viewed a body of water: as an opportunity to admire the reflection of his own self-image.

The cost of this narcissistic approach to controlling and changing place names is that it obscures how these names should be shared symbolic resources that serve the broader public and not just the ambitions, grievances, or political feuds of a single individual.

The conventional process of naming geographical features along the U.S.-Canada border involves bringing a formal proposal to the U.S. Board on Geographic Names and the Geographical Names Board of Canada, along with their state and provincial counterparts on both sides of the border. There is generally an extensive consultation process to ensure all geographical naming authorities are on the same page before moving forward with a name change. Yet Trump has thrown all of this out the window simply to rage-bait a U.S. ally into submission.

While the standard place naming process is not perfect, it at least establishes an institutional procedure for review, consultation, evidence gathering, and consideration of competing claims before a geographic name is changed. By contrast, what we are witnessing under Trump 2.0 is a unilateral naming by decree or whim with seemingly no public discussion or real government consideration of impacts and consequences. 

It is clear to us that Trump’s renaming of Lake Ontario is symbolic political theater aimed at diverting our attention away from the fallout of his trade war with Canada. Yet symbols matter. Place naming is a world-making practice that shapes the very foundations of our geographical imaginations and how we come to know and engage with the world.

There is a risk in dismissing Trump’s cartography of mass distraction as a frivolous symbolic tactic, since it can have real consequences for what is taught in classrooms, reconciliation efforts with Indigenous peoples, and the badly needed repair of relations between the United States and Canada.

If there is any hope to be found in the cartographic carnage that Trump is unleashing on the world map, it is that a future president can just as easily reverse these name changes in an attempt to restore America’s world image.

Recently, three events related to the U.S. Treasury bond markets have drawn public attention. For one, Japan sold some of its U.S. Treasury holdings to support the yen, and Treasury Secretary Scott Bessent intervened in the currency/debt markets to negate some of the resulting market pressures. Second, U.S. bond yields, especially at the long end, have risen alongside dollar weakness under the weight of an increased supply of dollar debt and weakening demand for it. And third, Secretary Bessent announced that the Treasury will purchase U.S. bonds, though its capacity to do so is limited.  While most people are inclined to view these as passing events, they are symptoms of a serious debt problem that appears to be progressing into a more advanced stage. 

In my book How Countries Go Broke: The Big Cycle, I laid out a template for understanding what happens when a country continuously spends more than it takes in, accumulating debt and debt service payments that rise relative to incomes. My perspective is that of an experienced global macro investor, and my understanding of this dynamic, which I will now explain, was what led me to anticipate the 2008 Great Financial Crisis and the European debt crisis that followed. 

Because I am now at a stage in life in which I want to pass along what I have learned, I wrote the book and am sharing this article in the hope that it will help people and policymakers deal with this issue well. 

How the mechanics work 

The debt dynamics of governments are analogous to those of individuals and companies with two important differences. First, when the demand for debt falls short of the supply, governments can create money through their central banks and hand it out to make it easier to pay debt (which also lowers the value of their money). Second, governments can get money from others through taxes. 

Throughout history, governments have tended to accumulate more and more debt until one or both of the following classic big debt cycle dynamics occur, leading to bad returns of debt assets and financial market crises.

First, debt service payments grow relative to incomes until they crowd out spending. Think of credit as being like blood in the economy's circulatory system. When credit circulates well and is used productively, it generates income that can service the debt that created it, which is healthy. But when debt service grows faster than the income needed to pay for it, debt-service costs accumulate like plaque in arteries, gradually crowding out other spending until eventually there is a financial heart attack. That is now happening in the U.S., but the U.S. isn’t alone. The United Kingdom, the European Union, China, and Japan all face too much debt relative to income and fiscal imbalances their governments haven’t solved.

Second, the supply of debt to be sold becomes much greater than the demand for it. As a country’s debt-service burdens grow, those who already hold a lot of the debt become less willing to buy the large amounts of new debt being offered and/or to roll over their maturing debt. Some holders even become more inclined to sell their debt holdings. At the same time, in cases when there are big capital and trade wars resulting from big geopolitical conflicts, history shows that fears of financial sanctions can hurt demand. When demand falls while supply is rapidly increasing, that causes interest rates to rise and credit growth to be curtailed. Credit and economic problems also typically occur, and there is often the "printing " of money and credit, which devalues it. In either case, that is bad for bonds. All these things are now occurring for the U.S., which adversely affects its government supply-demand balance.

The Big Debt Cycle degenerative process that follows these dynamics can easily be seen and understood by studying historic cases across many countries and is as predictable as demographic changes. Anyone who has studied monetary history should know that all monetary orders have eventually broken down, and it was this dynamic that led to the declines of previous reserve currencies and the empires behind them, most recently the British and, before that, the Dutch. Yet the process is not well understood and typically ignored until it is too late because, like unhealthy practices such as smoking and eating fatty foods, it takes place over a long time—typically over about a lifetime of around 80 years. The exact timing of the financial/economic heart attack is not easy to predict until the final symptoms appear. In my 2025 book, I estimated that it would take place in 2027, give or take two years. So far, the progression has been consistent with my estimates. It is certainly time to understand and pay attention to these dynamics. 

Just like the progression of symptoms with a disease, this degenerative process is observable and measurable. Rising debt burdens, weakening debt demand, increasing monetization, and deteriorating central-bank balance sheets can all be used as indicators of where in the cycle the process is and what is likely to happen. 

More specifically, the key red-flag indicators to watch out for are:

1. Government debt-service costs rising relative to government revenue to unacceptably squeeze out spending. 

2. The supply of government debt becoming too large relative to demand for it, causing long-term interest rates to rise faster than short-term rates.

3. The government treasury shortening the maturity of its debt sales to reduce the supply of bond sales.

4. The currency weakening, particularly relative to hard asset storeholds of wealth such as gold.

5. With a further lag, higher interest rates hurting the prices of other investment assets like stocks and real estate, and, after another lag, hurting the economy and creating credit problems.

6. Central banks "printing" money and credit, purchasing bonds, and guaranteeing debt. 

7. Central Banks incurring large losses and monetizing their own debt. 

8. Late in the cycle, governments adopting more extraordinary measures to manage the growing mismatch between their debt offering and debt service obligations and their available financing. These measures can take the form of: shutting down banks or forcing bank mergers because the banks' losses and lack of liquid funds make fully paying their depositors' withdrawals impossible; unusual financial supports for systemically important companies; the establishment of capital controls to prevent money from leaving the country; and the outlawing of hard asset monies such as gold.   

9. The process reaches a breaking point when debt service crowds out essential spending, bond supply overwhelms demand and pushes interest rates higher, or central-bank money creation becomes excessive and undermines the value of the currency. 

In all these scenarios, bondholders do poorly until debt and currency values are devalued enough to restore demand or the debt is restructured. Quite often these cycles end with a return to hard currencies and hard monetary policies to reestablish confidence in debt as an attractive, real-returning asset.  

That is the typical Big Debt Cycle. 

The U.S. situation in a nutshell 

To understand the U.S. position today, imagine that you are running a big business called the U.S. government.

This year, revenue will be roughly $5.5 trillion, while expenditures will be approximately $7.5 trillion, resulting in an anticipated deficit of nearly $2 trillion. Spending therefore exceeds income by roughly 40%. At the same time, federal debt held by the public is approximately $32 trillion, or about six times annual revenue and $240,000 per American household. Interest expenses alone are approaching $1 trillion per year, roughly 20% of revenue and about half the annual deficit.

In addition to interest payments, roughly $10 trillion of maturing principal must be refinanced. As a result, total debt-service requirements today amount to roughly $11 trillion, or about twice annual revenue.

That’s the current situation.

Looking forward, it appears most likely that things will get worse, and I estimate that projected deficits will cause the federal debt to rise to roughly $55 to $60 trillion over the next decade, requiring an additional $25-$30 trillion of debt sales. If that occurs, debt-service burdens will continue rising while increasing pressure is placed on investors to absorb ever-larger supplies of government debt assets. In addition, similarly large increases in debt and equity in supply in the U.S. private sector and in other countries that have to fund their increasing military and other expenditures will greatly add to the overall supply of debt and other financial assets.   

My 3% three-part solution

The proposed solution that I laid out in my book was to stabilize the government's debt and debt ratio to roughly 3% of GDP through a balanced combination of spending restraint, increased tax revenue, and lowered real interest rates (which would happen naturally with improved debt fundamentals). All three changes are necessary because relying excessively on any one of them would create severe and unnecessary economic pain because the adjustment would be too great. Based on my analysis, spending reductions and revenue increases of roughly 5% relative to current plans, combined with interest rates approximately 1% to 1.5% points lower than otherwise expected, would substantially reduce future debt-service costs from current projections. Lower financing costs, stronger asset prices, and improved economic activity would also support government revenues.

History shows that this kind of solution is possible. The most comparable U.S. example occurred between 1991 and 1998, when the budget deficit was reduced by roughly 5% of GDP while economic outcomes remained favorable. But because of the lack of dealing with this debt issue earlier and the resulting current level of indebtedness, plus the increased needs for capital to fund AI and military expenses, we may be past the point of no return. 

What this means for investors 

For investors, the lesson is not to try to predict the exact timing of the next debt crisis. Timing such events is a challenge for even the most experienced investors. Instead, what’s important is to recognize the long-term implications of excessive indebtedness and to diversify broadly across countries and asset classes, favoring balance-sheet strength, and be cautious about concentrating heavily in long-duration debt assets. Assets that are not government liabilities, such as gold, can provide useful diversification when governments are monetizing debts and depreciating their currencies.

Even though these debt dynamics have occurred repeatedly throughout history and are logical, they still surprise people. The key is recognizing them early enough to act before they become unmanageable. The warning signs are measurable, and the necessary adjustments are obvious based on the lessons of history. The question is whether political leaders and policymakers will understand this and act while they still can—and whether you and others will protect yourselves if they don't act. 

Data center companies “dug their own grave,” Texas Governor Greg Abbott said Sunday, and “that’s why they got the backlash they deserve.” Nine months ago, Abbott crowned Texas the epicenter of AI development, alongside Google executives announcing a $40 billion investment in the state. 
Pennsylvania’s Josh Shapiro, a Democrat who actively recruited Amazon’s $20 billion commitment to his commonwealth, made a similar U-turn days earlier, signing an executive order he calls the “nation’s strictest guardrails” on data centers: no more fast-track permits, no nondisclosure agreements, and no state permit review until developers make binding commitments to community standards.

When a Texas Republican and a Pennsylvania Democrat pivot to the same position in the same week, it becomes clear that the politics of data centers have shifted. Like everything else in AI, sentiment is moving at an extraordinary pace. Even the AI industry’s biggest winner recognizes the need for change. The tech sector needs to “do a much better job working with the communities,” Nvidia CEO Jensen Huang said this week. 

While shifting or walking back public statements never looks good, leaders are simply listening to voters’ demands. The truth is that leaders across the political spectrum are pivoting on data centers for a simple reason: they want their communities to actually benefit. The best ones are listening to citizens’ concerns while constructively raising the bar for the data center industry—not cutting them off. 

The costs and benefits of data centers

Seven in ten Americans tell Gallup they oppose a data center in their own community, more opposition than a nuclear plant draws. The primary complaint started with electric bills. PJM’s independent market monitor attributes 63% of the region’s 2025/2026 capacity price increase to data center load, roughly $9.3 billion recovered from customers across 13 states in a single year, causing household bills to rise 1.5% to 5% this summer.
The countervailing benefits are just as measurable. In Loudoun County, Virginia (often referred to as known as the "Data Center Capital of the World”), data centers supply as much as 31% of local revenue by the state legislature’s audit count, and the county has banked a $119.7 million stabilization fund against any downturn. The facilities also pay blue-collar wage premiums of 10% to 64% over comparable employers, the venture firm Andreessen Horowitz finds in Indeed postings data. The Dallas Fed has reported that skilled concrete workers, who typically make $28 to $32 per hour, are earning $45 per hour and a $150 per diem on data center jobs. And in July, the National Federation of Independent Business found that small-business optimism is at its highest levels in a year, crediting AI investment in chips and the structures that house them with spillover business opportunities.

The fight over data centers, meanwhile, has become a social and political flashpoint and, at times, embellished beyond what the record supports. Some backlash is earned. Most jobs promised are temporary. Resource demands are massive, and draining when mismanaged. Speculators have gamed interconnection queues and incentive programs, while both developers and policymakers have done a poor job communicating to the public. We heard these concerns from over 50 mayors leading cities that span the U.S. during our annual Yale Mayors College in March. 

Done well, though, these projects are a once-in-a-generation development opportunity for places ranging from underused farmland to urban brownfield sites. As Huang explained it: “This is the first time in probably the last century that we’re able to invest in sustainable energy, invest in improving our energy grid, securing our energy supply, reducing the cost of energy across the country.”

Local and federal responses

Opposition to data centers tends to concentrate at the county level because costs pool locally while benefits disperse. A county absorbs the water draw, the noise, the land conversion, and a share of the regional capacity bill. The compute serves users elsewhere, and the returns accrue to shareholders nowhere nearby. Every serious policy response attempts to move some portion of that diffuse benefit back to the community bearing the concentrated cost.

The responses span a diverse spectrum. New York paused state environmental permitting on large data centers for at least one year, pending a statewide study—the only statewide pause in effect, for now. Texas ordered an audit of every project in its grid queue, and the state’s grid operator paused new data center connections until the audit is complete. Maine’s legislature passed the first statewide construction moratorium this spring, only for Governor Janet Mills to veto the bill and create an advisory council to write standards instead. More broadly, 27 states advanced large-load legislation this year, with California, Ohio, and Utah enacting laws that exceed the industry’s voluntary ratepayer pledge to the White House.

 At the federal level, Washington continues to accelerate. A July 2025 executive order fast-tracked federal permitting, and a December preemption order, whose promised list of invalid state laws has yet to appear, expressly preserved state authority over “AI compute and data center infrastructure, other than generally applicable permitting reforms.”

Beneath the surface, the two sides are searching for a shared premise: the buildout must take the form of a mutually beneficial, transparent partnership. The public is already there. In Morning Consult polling commissioned for our June CEO Summit, 55% of Americans view data centers as a community benefit, while 69% call the backlash justified, the clearest indicator of an electorate searching for better terms, not fewer projects. 

“Don’t ban—raise the bar, ask for more,” suggests Chris Crosby, CEO of Compass Datacenters.

Meta’s Mark Zuckerberg drew the same position in his August letter, proposing a community compact of local jobs, investment in schools, and public services, held-down energy prices, and environmental care, backed by a $1 billion fund. When the company told Eagle Mountain, Utah, back in 2018 that it wanted to be a welcomed corporate partner, local public leaders answered, “Well, then we’re going to ask for some things.” And ask they did.

Virginia Governor Abigail Spanberger, whose state hosts more data centers than any other, has taken note and come to define the state’s role as ensuring “our local communities know what they can ask for and what standards they should be setting.” If knowing what to ask for is half the problem, then capacity to execute is the other half. The table often seats a county planning staff of a dozen across from a trillion-dollar counterparty, and that imbalance has unfortunately appeared in lopsided deals negotiated behind nondisclosure agreements.

Yet the smaller side of the table holds more leverage today than it may realize. The industry is desperate to scale and willing to deal. “AI is constrained today by data center capacity, not compute or software,” MARA Holdings CEO Fred Thiel tells us.

In reviewing the tariffs, statutes, audits, agreements, and published commitments behind the American AI buildout and other noteworthy economic development projects for our forthcoming book When Machines Act, we identified those practices that have produced win-win outcomes for all parties. 

Pay full freight, and sign for the total 

Our research suggests that major industry players have already accepted the central demand from communities. Data centers should pay the full grid costs of their own growth, and industry committed to do so when seven of them, from Amazon to xAI, signed the White House ratepayer pledge in March. Anthropic, though not among the signatories, also committed weeks earlier to cover the electricity price increases, transmission lines, and substations its own data centers require. “This should be the expectation across the industry,” the company wrote. Microsoft went even further, telling regulators to set rates high enough so that data center costs do not impact anyone else.

A pledge is only as good as the paper behind it, which is why Google’s Capacity Commitment Framework matters as much as any of them—a structure that includes guaranteed minimums, a security deposit, and cancellation fees a utility can collect. Water has followed the same path from promise to specification. Microsoft is aiming for a 40% improvement in datacenter water-use intensity by 2030, and Meta has pledged to restore double what it consumes in stressed watersheds.

Flexible demand response is another potential lever few contracts have adopted and is one of the cheapest options available. Duke’s Nicholas Institute calculates that if data centers trimmed their peak demand just 1% to 2%, retail electricity prices could fall as much as 2.8%. Utilities build power plants and wires to cover the few highest-demand hours of the year, so a data center that briefly tempers or shifts computing during those hours reduces the amount of new generation and transmission that must be built. The avoided construction flows through regulated rates as savings to every customer. Operators can avoid performance losses—shifting tasks in time or place during peak periods—and save up to $8.00 per megawatt-hour for their contribution.  

Trade tax breaks for assets that last 

Incentives are a different matter, costing taxpayers real money, so the first question for any state is whether a subsidy changes a company’s decision. Georgia’s independent state audit found that the sales-tax exemption changed the siting decision for roughly 30% of data center activity, and while the incentive produced a net fiscal loss of $433 million in fiscal 2025, it also returned $2.86 in economic value across the broader economy per dollar forgone. While critics carry the loss and boosters the multiplier, the audit’s real lesson belongs to neither camp. A state that pays for 10 projects to decide only three should be asking for far more than it does.

A study by the Brookings Institution finds that hyperscale sitings follow power, land, and fiber, meaning subsidies largely pay for what was coming anyway. The smarter deal asks the incoming project to fund assets the community keeps, such as transmission upgrades, roads, and water systems, that compound in value over the long run. 

Compass Datacenters’ Crosby volunteers that rationale himself. On a $5 billion project, spending $100 million on transmission upgrades “makes a lot of sense to me from a real estate perspective.” Taking a page from semiconductor industrial policy—such as Micron conditioning its Central New York megafab on a $500 million community fund shaped by 300 local civic groups—states can require data center operators to seed long-term local endowments. And New Jersey is conditioning AI data center credits on partnerships with in-state universities and startups, converting a tax break into innovation-economy seed capital. 

Write the rules before the applications arrive 

Seed capital compounds only where the rules were written before the arrival of the applications. Eagle Mountain set expectations first and approvals second. Meta spent roughly $100 million on roads and electrical service before construction, backed a creek restoration project the city expects to return 476 million gallons of water a year, and granted more than $1 million to local schools and nonprofits. The company is carrying that playbook forward to Hyperion, its largest campus yet, in Richland Parish, Louisiana, where it has pledged more than $1 billion for local roads, water, and wastewater systems and funded data center trade scholarships for every parish high school graduate—and where surging parish tax receipts funded $50,000 teacher bonuses this year.

Yet both campuses also carried confidentiality and nondisclosure agreements that cast a shadow over the goodwill earned from community investments. Data centers need to deal transparently and operate openly, since closed-door agreements breed suspicion even when the terms are generous.

Public verification, therefore, matters as much as a pledge to a long-term partnership. The city of Lancaster, Pennsylvania, publicly posted its full agreement with a data center consortium, secured the obligations with a letter of credit, and bound the counterparties to the land, all while granting no city tax breaks. Georgia’s audit process exists because of a requirement from a 2024 transparency statute. New York attaches a uniform per-megawatt host fee to large renewable projects—a renewable-siting design that data center statutes could borrow—and researchers studying the first twenty projects under the law found the fee became a floor that communities negotiated above. From another parallel case, Massachusetts addresses the root of the asymmetry by allowing the gaming commission to require casino applicants to pay the host community’s legal expenses. In policy, the standards embraced by Spanberger provide communities with the knowledge of what to ask for and how to ask for it well.

How a state sequences these tools matters as much as which it selects. States must build the accountability layer first, requiring disclosure, published agreements, and scheduled audits before offering a single incentive. The negotiated package of local benefits that Lancaster and Eagle Mountain modeled (host community frameworks) and the rate design that binds a data center to the full grid costs the industry has pledged (cost-coverage tariffs) should then pair local benefits and ratepayer protection as one package. Any incentive that follows needs to be conditioned on a “but-for test” fortified with a proportional clawback instrument, and only then should states pilot novel concepts such as flexible demand response and ownership stakes that give host communities equity in the projects themselves.

An effective state framework

Ultimately, we have found that four universal principles govern an effective state framework:

• Commitment prior to permitting: Binding community and grid compacts must precede site approvals.

• Proportionality: Infrastructure and community asks must scale directly with peak megawatt demand.

• Enforceable security: Pledges must be backed by letters of credit, escrow reserves, or parent-company guarantees.

• Statewide baseline floors: Establish statewide statutory floors to prevent developers from regulatory arbitrage across county lines.

A year ago, data center developers and their backers spent little to no time on policy. Today, some, Crosby among them, are spending as much as 80% of their time on it. As he put it: “Our social license is at risk, and we have to figure that out as an industry.”

Willingness, at times, has even outrun the law’s ability to receive it. Compass Datacenters offered the Texas Public Utility Commission $100 million toward transmission upgrades but was refused because existing law did not permit the entity to accept the funding. The mismatch captures the moment. The moment is moving faster than the rules that govern, and that pace obliges data center owners and developers to work closely and transparently with the states, localities, and communities that host their projects.

When managed properly, we believe these facilities can be monumental opportunities for the towns that attract them. The buildout, in Huang’s words, “is going to re-industrialize the United States.” If the trend persists, however, the moment could be lost to neighboring nations that do embrace them, such as border towns like Tijuana, Mexico. Even more, the demand underneath them carries national weight, too, as the buildout will help decide whether America leads the AI race with China or falls behind. What remains is establishing enforceable terms that let both sides meet in the middle, so the communities hosting the infrastructure of the AI era also hold a durable share of what is built in their backyards.

In 1937, Aldous Huxley ominously warned in Ends and Means: “Technological progress has merely provided us with more efficient means for going backwards.” As communities across the nation and across parties come to embrace this neo-Luddite resistance to data centers, some developers are listening, learning, and hopefully acting.

With research assistance from Frankie Reichman and Zander Jeinthanuttkanont.

I don’t expect anything of my friends that I don’t expect of myself.

But lately, I’ve found myself disproportionately annoyed by tiny breaches of friendship etiquette. If I ask a friend to dinner and they say they’re busy without suggesting another date, I feel slighted. If I introduce two friends and they start hanging out without me, I wonder why neither thought to include me. Then I wonder: am I holding my friends to an outdated social code? 

Today, it is common to say “let’s catch up soon” without making plans, leave invitations unanswered, and allow one person to become the permanent initiator. My theory: Technology lets us remain constantly in touch while making the obligations of friendship increasingly ambiguous.

When you think about it, an invitation is more than a logistical question, says Dr. Jeff Katzman, a psychiatrist at Silver Hill Hospital in New Canaan, Conn., who extensively studies human relationships. It’s a small relational bid: something close to, “I’d like to spend time with you. Would you like to spend time with me?” When that second part is missing, we’re left with ambiguity, and we fill in the blanks. 

“From an attachment perspective, we’re continually looking for signals about whether the people we care about are available and responsive,” says Katzman. “When I reach toward you, is somebody there? We also bring our own histories to these moments.”

Someone who has experienced rejection or exclusion may experience the same unanswered invitation quite differently from someone who has generally experienced other people as reliable. Katzman points out that we do our best to infer other people’s minds from very small pieces of behavior. “A text goes unanswered, and the human mind is remarkably good at writing the rest of the story: ‘She doesn't really care about me. He doesn’t want to see me. I’m always the one who tries,’” he explains. 

I have lost a lot of sleep about the idea of social hygiene and the small acts of reciprocity that keep friendships healthy. I was quite surprised when two of my friends in Lisbon reached out to me on the exact same day when I had said to them separately a few months ago that I might be visiting Lisbon for work on that date. The fact that both of them remembered the exact date and reached out to check if I’m in Portugal is a great example of good social hygiene. 

Whenever I make new friends now, I pay close attention. Do they regularly cancel plans at the last minute? Do they have the capacity for investing in a friendship and take on the logistical requirements of keeping an adult friendship alive? It takes roughly 50 hours of time together to move from mere acquaintance to casual friend, 90 hours to go from that stage to simple friend status, and more than 200 hours before you can consider someone your close friend, according to a report published in the Journal of Social and Personal Relationships. Getting to know someone takes a real commitment of time. 

I also think about rejection and what we reasonably owe our friends. “Reciprocity is extremely important, but I don't think reciprocity means symmetry,” says Katzman. Healthy friends don’t need to keep score. “I think of friendship a little like improvisational theater.” One person makes an offer, and the other person receives it and makes an offer back. The contributions don’t have to be identical. They’re building a scene together. One person might initiate more dinners; another might get curious a little more reliably. And at different stages of life—parenting, illness, caregiving, grief, work pressures—one person may carry more of the relationship for a while.

In good improvisation, both people help create the scene. Friendship is similar. We don’t have to say “Yes” to Tuesday night, but somehow, over time, we need to say “Yes” to the relationship. And it’s best if we can let our friends know that, in some way.

“A useful question might be: if I stopped doing all the work of maintaining this friendship, would a friendship still exist?,” asks Dr. Katzman. Sometimes, it may be worth stepping back a bit to see what happens.

When I have done that in the past, a lot of the friendships naturally dissipated, whereas when I look at all the close friendships I have now, it’s the result of mutual effort. It’s pretty similar to playing tennis, where one person hits the ball, and the other person hits it back. Sometimes, one of us drops the ball, but quickly enough it’s picked back up again to continue the game.

At the same time, we should have some humility about how many relationships any person can actively maintain. Former U.S. Surgeon General Dr. Vivek H. Murthy, in his book Together, describes concentric circles of connection: an intimate circle of close friends and confidantes, a larger relational circle of friends and companions, and a much larger collective circle of colleagues, acquaintances, and community. We need all of these kinds of connections, but we can’t maintain hundreds of relationships with the intensity of our closest friendships.

So I took a fountain pen and drew concentric circles in my journal and assigned a circle to every friend, be it close friends or acquaintances. It helped me understand that someone can genuinely value our relationship without having the bandwidth to place us in their innermost circle at a particular point in life. It also helped me assign how much energy I want to spend.

Ultimately, it helped me appreciate my friends for what they are.

The expansion of illegal Israeli settlements and outposts in the West Bank, alongside the ongoing horror of life in Gaza, makes the need for a two-state solution between Israel and Palestine more urgent and necessary than ever. International dithering and procrastination must stop to preserve any chance of sustainable peace.

This was our conclusion after visiting the West Bank as part of a wider Middle East trip that included meetings with Palestinian, Israeli, Lebanese, and Jordanian leaders. 

It is clear to us that the Government of Prime Minister Benjamin Netanyahu’s policies threaten to make Palestine disappear politically, culturally, economically, and physically. 

Traveling through the West Bank, we had a visceral sense of a land and people under deepening occupation. Everyday life is impeded at every turn. Settler violence is an integral part of this strategy, designed to intimidate and terrorize Palestinians off their land. Far from being the work of a small number of “juvenile delinquents” as he claims, it is a state-backed campaign undertaken with the active support of the Israel Defense Forces (IDF). 

We saw at first-hand how state-backed settlement expansion operates. First, hillside outposts are established by a handful of settlers. Even though these are illegal under Israel’s own laws, we heard directly how they are systematically and swiftly protected by the IDF despite Israeli authorities from time to time condemning settler violence. Connections to water and electricity supplies typically follow, as do roads reserved for Israelis only. 

The result is the physical and administrative carve-up of the West Bank. Palestinian communities are distanced from each other, their land, and the possibility of a future Palestinian state. 

Among the blatant evidence of what is happening is the decades-old E1 settlement plan for Palestinian territory east of Jerusalem. On Aug. 18, the Israeli Housing Ministry issued a tender for over 1,200 units, setting off a process which,  if completed, would sever East Jerusalem from the rest of the West Bank and thereby disrupt the territorial contiguity of a Palestinian state.

E1 has long been a red line even for Israel's staunchest allies. Previous Israeli governments, including those led by Prime Minister Netanyahu, have stopped short of actually proceeding, in part due to international pressure. Now that commercial tenders are being issued and construction contracts potentially signed, any companies and banks tempted to get involved would be breaching international law and have been warned to that effect by a number of Western countries and the European Commission.

Israel’s friends and allies must demand E1’s cancellation—not just postponement or freezing, but definitively ruling it out as a threat to the two-state solution which is necessary for both Israel and Palestine to have peace and security. 

Many Israelis recognize this threat and oppose the growing extremism and intolerance in their politics and society at large. For instance, we met with President Isaac Herzog on our visit, who condemned settler violence. We also acknowledge the deep trauma Israel’s people suffered after the Hamas terror attacks of October 7, which included the detention of hostages contrary to all international law. 

We take inspiration from the brave Israeli and Palestinian civil society activists we met. They remain determined to combat impunity for abuses and injustice, and to strive for a future for both peoples based on equality and respect for international law.

Yet this vision is far from the reality we saw on our visit. We saw how ongoing settlement of the West Bank and the sustained violence against and displacement of Palestinians threatens ethnic cleansing and annexation. Combined with the occupation of Gaza, including renewed calls by Israeli ministers for new settlements in the Strip, we are convinced that what is underway is the conquest and erasure of one state by another. 

This is the logic of some of the bloodiest wars of previous centuries. It is what Russia is seeking to do in its illegal war of aggression against Ukraine. But it cannot be an acceptable way for a country which calls itself a democracy to behave in the 21st century. 

Our visit ended in Brussels, where we met with representatives of the European Commission and EU member states. The EU was founded as a values-based organization to promote peaceful cooperation between nations. Given its significant trade relations with Israel, it could play a decisive role now in exerting pressure on Prime Minister Netanyahu and his government.

Yet what is lacking is the political will to take even basic steps. In line with international law, the EU considers Israeli settlements to be illegal, yet EU members continue to trade with Israel. Suspending preferential trade under the EU-Israel Association Agreement is another overdue step due to Israel’s failure to comply with human rights requirements, as already called for by the European Commission. Nor has the EU collectively sanctioned extremist ministers Smotrich and Ben Gvir. 

There is still time to change course before it is too late, but only just.  

When Tiangong Ultra unofficially smashed Usain Bolt’s 100m world record in Beijing, China, unballetically flailing past the finish line in 8.86 seconds, he did not flash an Olympian grin down the camera lens. In fact, he has no face. He is a headless robot.

Ultra, the work of the Beijing Humanoid Robot Innovation Center, was competing in the penultimate day of the unapologetically dystopian World Humanoid Robot Games, which has been hosting 666 teams at China’s National Speed Skating Oval. At this event, more than 2,000 robots are put to the test in track and field events (like long jump and an obstacle course); sports (like table tennis and soccer); and “scenarios” (like bricklaying and firefighting). 

The event produced several eye-catching moments, including when humanoids beat human records at the high jump and the 400m. Just a few days before Ultra’s race, Lightning, a robot developed by phone company Honor, ran the 100m in 9.39 seconds, again thundering home quicker than Bolt.

But these tumbling records will likely be, in the pop cultural consciousness, eclipsed by something far more entertaining: tumbling robots. Along with the epic wins, there have been some seriously old-school, epic fails. One humanoid at the weightlifting event lost its balance with a weedy 15kg barbell, started jerking and slammed into the judges’ table, its helpless arms aloft, as if to say, “Why me?” Most memorably, a robot ended his dash by careering into a safety mat, Tom and Jerry style, before cartoonishly arching backward and setting on fire. Each spark was like a beautiful firework.

These slapstick scenes are deliciously satisfying. It’s deeply reassuring to watch robots fizzle out into smithereens. I myself have fond memories of watching the fire-hazard creations on Robot Wars in the U.K. (the successor to BattleBots in the U.S.) get shredded into shrapnel. This time, it’s a reassuring reminder that we can beat robot replicants at our own games. 

This robot schadenfreude is fuelled by a fear of the future. Pointing out their flaws makes humanity’s finish line feel further away, even if that’s a highly unintelligent thing to think.

In many ways, this feeling can be filed next to our twisted delight in belittling generative AI. ChatGPT’s past inability to spell “strawberry” has made it an object of ridicule online. Calling AI-generated content “slop” has become a knee-jerk reaction to artificial intelligence’s catastrophic metastasisation in the creative industries. 

And our collective giddiness about technical mishaps is not reserved just for AI. For instance, we love to watch the armored glass windows of the Tesla Cybertruck break. One possible reason: it shatters the illusion that the future is here and already decided. Robots surely can’t replace us if we haven’t figured out car windows yet. This feeling is magnified when they are humanoids—ersatz versions of ourselves. 

Seen through a sports lens, it mirrors how I felt with the Enhanced Games. The painfully brotastic event held in Las Vegas in May allowed athletes to take performance-enhancing drugs or wear banned gear. With social media instilling a chronic fatigue for actual human achievement, technophiles are moving the goalposts; if being pumped full of steroids or titanium plates makes you more likely to beat a world record, it’s worth it. Once again though, disappointment was on show. Just a single best-ever-time, in the men’s 50m freestyle, was (unofficially) recorded by Kristian Gkolovmeev; he wore an outlawed polyurethane skinsuit. Otherwise, the non-artificially-enhanced athletes dominated the competition.

My concern: when it comes to humanoids, at least, our schadenfreude is set to be a short-lived pastime. While advanced robots are still relatively primitive, their improvement is staggering. To put it into perspective, at last year’s World Humanoid Robot Games, Ultra won the 100 meters in 21.50 seconds, much closer to Bolt’s 200m record than his 100m time. They’re only getting stronger and faster. Dunking on “clankers,” a derogatory term for clumsy robots (like this one dancing to Billie Jean), may be the most classically human response: laughing off the impending doom. 

Many futurists predict that in the near future, our factories and farms and battlefields and back gardens will be filled with humanoid helpers. In 2024, Elon Musk said that by 2040 there will probably be more humanoid robots than people. And according to research by Morgan Stanley over 1 billion humanoids may be in use by 2050 and the market could be worth $5 trillion.

Spectators from afar like me, viewing the drama on our phones, will continue to watch with a sort of futuristic gallows humour. ​​Laughing at misfortune, however spiteful it may be, leads to social bonding—in this case, as an entire species, against robots—that humans desperately need right now.

Every time a clanker goes crash, bang, wallop, and kaput, it’s a visceral reminder that we still matter. And unlike them, when faced with their destruction, I’m able to smile.

At a recent academic retreat I attended, the air was thick with what I can only call educational gaslighting. A panel of graduate and undergraduate students looked a room full of professors in the eye and claimed they only used AI to verify their work because they valued learning too much to take shortcuts. Minutes later, when the answers were blind, those same students estimated that over 80% of their peers were using the technology for nearly everything.

As an engineering professor at the University of Michigan, I believe we need to move past the fear and hype. The future job market will not be dominated by autonomous AI, but by experts who have mastered their field so thoroughly that they can use it to multiply their output exponentially. 

But how do we help students become experts if they don’t show up? 

This question precedes the LLM onslaught. Since the pandemic, traditional lecture attendance has cratered, but active learning has been shown to significantly improve both turnout and long-term retention. 

By evolving my courses to embrace this data, I’ve seen attendance surge, even in the freeze of a Michigan winter. 

Flipping the lecture cycle

Classically, engineering courses default to hours of lectures where students are expected to take notes, with problem sets and exams bolted on. Many students treat a lecture as passive entertainment. And, often, the material is so technical that it is disconnected from real-world use, leading to even less engagement and retention. 

To break this cycle, I’ve flipped my classroom. Each week, I assign a 2-hour recorded video lecture, along with a related article. The assignments are made in Perusall, an AI-enabled tool that treats the video and article a bit like a social network. Students are graded based on their active engagement with the material, such as how much of the lecture they view, what questions and comments they leave in the system, and so on. I can monitor which students leave comments, answer peer questions, and engage with the material before they ever set foot in my classroom. And if they try to cut and paste comments in multiple locations, the system flags them. It does not yet flag comments that seem AI-generated, but I expect that will be coming soon.

With everyone primed to dig in, only one-third of my students’ time with me is devoted to classic lecturing. I offer a one-hour live lecture and invite industry guests to share stories of computer vision in the wild. Then my students spend the rest of our in-person time participating in small breakout sessions, a large group discussion, and an in-person quiz. Not only do they grade their own quizzes, but they only get credit for an answer if one of them argues the logic behind it.

The result? My students show up to class because the value is no longer in the information I provide—it's in the friction and growth of live exchange.

This fall, I’m taking this a step further. We won’t just read technical papers; we will debate them. Anyone can be called to the front of the room to spontaneously argue one side of a research argument, which means every student must come prepared.

By moving the passive learning to the home and continuously pushing students to test their knowledge, I’ve reclaimed the classroom to create what AI cannot replicate: spontaneous, high-stakes human interaction.

Using AI as a supercharged tutor

As we try to understand how AI can help and hinder learning, the most dangerous misconception is that it is a labor-saving device for the mind. In reality, AI is an expertise-amplifier that can turn weeks of manual programming into a few hours of focused work. But for a novice, relying on AI before mastering the fundamentals creates a technical debt that leads to a lack of depth.

As someone at the forefront of AI research and creation, I don’t coach my students to avoid it, but rather I use it as a sophisticated, one-on-one tutor that facilitates active learning and helps them grow their expertise. This means moving beyond passive consumption and toward a rigorous, iterative process of trial, error, and refinement.

Some best practices I share with my students include:

Mastery-First Workflow: Solve problems manually first. Then use AI to check your work and identify where your logic diverges from the model.

AI as a Problem-Generator: One of the most effective ways to learn is through constant testing. Use AI to generate new practice problems and engage in active learning.

Brain Dump Standard: Never ask AI to write from scratch. Instead, provide a brain dump of ideas and structure. After the AI helps organize your expertise, personally refine it through meticulous review or even rewrite, if necessary.

Redefining the honor code in the age of AI

I am not an AI police officer. I cannot—and should not—spend my academic career hunting for digital shortcuts in my students’ work. I can only set the boundaries and allow them to choose how they show up. 

Amid the promise of AI to supercharge the work of experts, we must treat this technology with the same proactive mastery we apply to any other essential tool of modern life. 

By shifting the focus to high-stakes, spontaneous human interaction and leveraging AI for active learning rather than trusting it to do the work, educators can ensure that the knowledge lives within the student, not just the model.

There is no epidemic of noncitizens voting in our elections. Federal law already bars it, and despite repeated allegations by a number of prominent conservative figures, hunts for violators have turned up next to nothing.

Yet President Donald Trump continues to aggressively push for passage of his SAVE Act, a bill that would require every American to show a passport or birth certificate in person just to register to vote, effectively ending online and mail-in registration. Framed as a fraud-prevention measure, I expect the bill would ultimately do far more to keep eligible citizens off voter rolls than to stop alleged noncitizen voting. And then there’s what Trump expects: he has said that the SAVE Act would “guarantee the midterms” for Republicans, who, he’s promised, “will not lose an election for 100 years.”

But the SAVE Act should scare voters across the political spectrum. Free and fair elections are the cornerstone of our democracy. They are the primary mechanism by which we select our leaders and hold them accountable. When any politician—whose access to power depends on election outcomes—repeatedly casts doubt on election integrity without evidence, we must see it as an attempt to undermine our democracy and push back hard.

This is why I see the passage of the SAVE Act as the equivalent of a four-alarm fire for our democracy. Not only could the bill make it impossible for millions of Americans to exercise their right to vote, but it would also hand the federal government unprecedented new powers over election administration and voter data, powers that have always belonged to the states. Such is the potential for mass disenfranchisement and abuses of power that even members of Trump’s own party have decried the bill as “a mistake.”

And yet the SAVE Act is only one front of the attack on our democracy. There are other examples of disturbing ways that the Trump Administration is fearmongering about immigrants, effectively sowing mistrust in our electoral systems.

Since May 2025, the Department of Justice has demanded that nearly all states and the District of Columbia turn over full, unredacted voter rolls, including driver’s license and partial Social Security numbers. When most of those states refused, DOJ filed lawsuits against 30 of them and D.C. For the states that did provide the data, DOJ then shared it with the Department of Homeland Security to supposedly “scrub aliens from voter rolls.”

Since then, the pressure has only escalated. In July, a day after Trump gave a primetime speech in which he again railed against immigrants, made unsubstantiated claims of noncitizen voting and demanded that states change their election policies, Homeland Security Sec. Markwayne Mullin then threatened state election officials with prison time if they don’t acquiesce to Trump’s demands.

The intimidation isn’t just limited to officials; it’s also aimed at voters. Trump’s longtime advisor Steve Bannon has openly called to “have ICE surround the polls” on Election Day and recently brushed off criticism of that proposal, saying that, “it should be intimidating.” When Trump himself was directly asked if he planned to deploy Immigration and Customs Enforcement (ICE) and Customs and Border Protection (CBP) to polling stations, he equivocated, stating that he would do “anything necessary to make sure we have honest elections.”

Both federal and state laws prohibit the deployment of federal agents at polling places. However, we have already seen what fear does to a community under ICE presence. If people in Minneapolis, Chicago, and Los Angeles were too afraid to venture out for basic needs like groceries or a doctor’s visit rather than risk an encounter, what should we expect with the looming threats of a militarized presence around elections? How many people will stay home rather than risk being racially profiled and brutalized?

It is ironic that this is all unfolding in the year of America’s 250th birthday. And it begs the question: Are the actions of this administration consistent with the basic tenets of the Declaration of Independence? It is clear to me that they are not. 

By demonizing immigrants, the Trump Administration has endangered due process, the balance of powers, and the Constitution. Now, it’s using similar fear tactics to come for the very thing that makes self-government possible at all: elections we can trust.

The stakes here outlast any single midterm. Whether our democracy survives this depends on our collective ability to recognize the pattern before it’s finished—and to stop it while we still can.

And it depends on our ability to identify if we actually have something to be afraid of. 

When governments ask people what truly matters to them, they cite things like being healthy, having a secure and affordable home, enough income to live with dignity, and time for the people they love. As the former leaders of Iceland and Scotland, we have heard concerns first-hand. 

But these people-first priorities are not abstract ambitions. They are the conditions for a decent life, and ensuring they are met should be the first priority of any government.

And yet none of this is captured in the number that most economists, the media, and indeed, governments use to measure a country’s success: Gross Domestic Product (GDP), commonly referred to as economic “growth.”

To be sure, GDP has its uses as a measure of many of the goods and services a country exchanges and produces. But today, it is typically used in a way its architects never intended. For instance, a country's GDP rises when a forest is felled and when an oil spill is cleaned up. It rises with financial speculation. But it does not rise when a parent cares for a child or a child for an elderly parent, when the forest is left standing, when a woman can walk home at night without fear, when voters turn out because they trust their institutions, or when a patient sees a doctor in days rather than months.

A new UN Framework, Beyond GDP, promises to change this by offering a dashboard of more meaningful indicators to complement GDP. Governments must now put this framework into action. Not simply by publishing new statistics, but by using them to set priorities, shape budgets, assess policies and be held to account publicly on whether people’s lives are genuinely improving.

Eight years ago, we set out to put the wellbeing of our people ahead of the narrow pursuit of economic growth. Together with New Zealand's Jacinda Ardern, we founded the Wellbeing Economy Governments partnership, later joined by Wales and Finland, with Canada actively participating. We shared practical lessons about developing national wellbeing indicators, embedding them in government decision-making and ensuring that economic policy considers social and environmental outcomes alongside financial ones.

In Iceland, we introduced 39 wellbeing indicators, built on what people themselves told us they valued, and used them to inform policy. GDP became one measure among many, alongside life expectancy, unmet healthcare needs, material deprivation and work-life balance. Parental leave shows what this means in practice. We extended leave to 12 months, with six months reserved for each parent and six weeks transferable. In doing so, we recognized that care, family life, gender equality, and the economy are inseparable, and that fathers' time with their children is worth protecting. The reform strengthened families and shifted expectations about work and who does the caring.

In Scotland, a Wellbeing Economy Monitor showed us how the economy was really performing for people, alongside a National Performance Framework of outcomes written into law and aligned with the Sustainable Development Goals. Policies followed: the doubling of early years education, a Baby Box to equalize children’s starts in life, the Scottish Child Payment to reduce child poverty directly, and Community Wealth Building to ensure public spending, land and assets create lasting value locally—an approach Scotland has since made the subject of the first national legislation of its kind anywhere in the world. 

While these efforts are still evolving and not perfect, they represent an important shift in how governments understand prosperity. Changes in economic systems take time, but progress begins by changing—and being more explicit about—what we value. Our experience taught us that measuring what matters is only the beginning: wellbeing indicators must also shape budgets, policy decisions and how governments are held accountable. Legal systems need to support this too.

We were, and we still are, far from alone in this endeavor. Bhutan, for example, has been making the argument for more than 50 years. Its Gross National Happiness framework focuses on nine different dimensions of well-being such as living standards, health, education, and ecological diversity. Last year, Malaysia introduced a bold new roadmap that embeds public health, environmental sustainability and economic resilience. The National Planetary Health Action Plan moves away from a narrow focus on economic growth to prioritize the health of people and planet, replacing the short-term idea of return on investment with a “return on values.”

It is worth noting that wellbeing economy approaches often mirror the ways many Indigenous Communities have been providing for collective needs for generations. For example, Buen Vivir—living well together—comes from the Quechua peoples of the Andes, and has been written into the constitutions of Ecuador and Bolivia since 2008. And Aotearoa New Zealand's wellbeing budget draws on Māori understandings of intergenerational wellbeing.

These ideas are finally moving from the margins into mainstream economic policy. A landmark UN report, written by a high-level expert group appointed by the UN Secretary-General after the Pact for the Future, proposes a dashboard of 31 indicators to sit alongside GDP, with recommendations for governments, business, academia, and civil society on how to bring this agenda to life.

Earlier this year, a roadmap for eradicating poverty without relying on endless growth was launched in Geneva, developed with more than four hundred contributors from governments, trade unions, social movements, UN agencies and universities. It sets out measures that already work in different places, from social protection and care; to tax, universal basic services, and the rights of nature; to the governance of trade, debt, and finance. That work and the work on measurement are currently proceeding separately. Bringing them together would be a valuable contribution of the intergovernmental process now underway. 

The greatest risk now is a process in which governments agree in principle and move slowly in practice. New indicators alone will not be enough. The pursuit of growth at any cost is embedded far beyond the statistics: in how credit agencies rate a country's debt, how finance ministries deem a budget responsible, and in the rarely-questioned assumption that more is always better.

What is needed is not further consensus but more early adopters: countries willing to pilot implementation and share honestly what they learn, supported by civil society, business and the UN itself.

The 2015 Sustainable Development Goals ask every country to sustain economic growth per person. As governments begin to design what will follow them, we have a chance to change the instruction and create a different approach.

After all, the responsibility of every leader is to ensure the wellbeing and safety of his or her people, and to do that, we need to go beyond GDP and think about what truly matters

Claude now watermarks AI-generated text to comply with European Union transparency rules. OpenAI and Google add invisible fingerprints to AI-generated images. And Substack is touting a feature that scans pieces for signs of AI. Will we finally be able to tell what’s real on the Internet? My take: not even close. 

In fact, AI watermarks and detectors may leave us worse off by creating a false sense of confidence in content marked as genuine.

Watermarks and detectors are gaining traction as we lose our ability to trust our senses online. Look up the Will Smith eating spaghetti test, and you’ll see just how far AI has come. A 2023 AI-generated video shows the actor slurping spaghetti, face distorted, in a way that breaks physics. By 2025, AI was producing lifelike renditions. Deepfakes are so good that experts recommend families develop secret codewords to identify one another. 

“But I know a fake when I see it,” someone might say. 

Unfortunately, research consistently shows that you do not. This can feel especially hard to accept given the abundance of AI slop rocketing around the Internet. You may even start to think you can sniff out offending content. It might work, for a little bit. It almost never lasts. Any signal that becomes discernible is one a sophisticated actor will find ways to avoid. 

We’ve seen this story before. During the earliest days of the Internet, visual polish at least told you something. Major institutions had the resources needed to produce well-designed websites. Janky-looking sites, on the other hand, screamed “scam!” Information experts directed Internet users to dwell on features such as design, broken links, and typos. But when the Internet changed, the advice didn’t. 

A study I led, published in 2022, found that 96% of America’s leading colleges and universities offered outdated advice on how to evaluate online information—long after platforms like Wix, Squarespace, and Photoshop made it easier for bad actors to create fake but convincing-looking websites. Inexpensive software made slick graphics ubiquitous. Educators, however, continued to instruct Internet users to search for visual clues like a game of Where’s Waldo?

The most dangerous legacy of this aesthetic fixation is the inverse illusion: the cognitive tendency to believe that if the presence of a signal proves one thing, its absence proves the opposite. Yes, a site with misspellings that claims to show aliens still isn’t legit. But a beautiful site with a dot-org domain can also be harmful. In 2019, our research group found that nearly half of hate groups had dot-org domains. Bad actors know how to adopt the trappings of credibility. 

The same is true with AI. Even if visible flaws sometimes linger, their absence doesn’t mean content is genuine. Yet, too often, experts offer surface-level clues to identifying AI-generated content. This is why in the lead-up to the 2024 elections, Stanford Professor Sam Wineburg and I warned about public officials who advised citizens to pay attention to lighting, strange shadows, or other visual cues to identify deepfakes, even after AI content stopped making these errors. Many 2026 guides to spotting AI content mislead readers with the same poor advice. 

Which brings us to AI watermarks and detectors. These approaches, based on hidden signals in content, promise that while we can’t always spot the signs, their algorithms can. 

I’m not a software engineer. Yet I was able to easily strip metadata from some AI-generated images just by screenshotting them. Anthropic confirms that file metadata can be “stripped through format conversion, re-saving, screenshots, or other means.” Watermarks like SynthID are stronger and can persist after screenshots. But I was able to use a free online tool to remove a SynthID watermark. 

Google admits that the accuracy of detecting watermarked AI text is “greatly reduced” when users thoroughly rewrite what they generate, and that it “is not designed to directly stop motivated adversaries from causing harm.” More broadly, open-weight AI models that can run locally, outside platform terms and conditions, guarantee the spread of unmarked content.

Third-party detectors, too, have a spotty track record. I’ve regularly run AI-generated text through detectors that said it was human and vice versa. Many studies of text, image, and audio detectors find that they don’t work very consistently, and yet, their findings are used as the basis for public accusations. Every detector must confront an arms race with humanizer tools and other workarounds motivated actors find. 

I would argue that the biggest problem for detectors and watermarks remains the inverse illusion. Just because content lacks a watermark doesn’t mean it wasn’t produced or edited with AI. As Anthropic notes: “lack of a detected mark doesn’t mean the content wasn’t AI-generated or processed.” Deferring judgment to AI detectors leaves us vulnerable to bad actors who know how to launder content and make it pass muster.

This is a confusing time. Many of us are, understandably, uncertain. In one recent pilot, our research group showed 117 students a confident chatbot answer about local history with hallucinated facts. Half said they weren’t sure if it was true. One student said AI is sometimes right and sometimes wrong and “you never know which is which.” 

But just because we can’t trust our eyes or place full faith in detectors doesn’t mean we can’t trust anything. Rather than hunt for visual clues or outsource judgment to detectors and watermarks, we can turn to reputation and context. It’s easy to fake content. It’s much harder to fake a good reputation that’s validated by credible sources. 

The next time you see unfamiliar content online, resist the urge to ask, “Does this look like AI?” or run the content through a detector. Instead, ask yourself, “Do I trust where this information is coming from?” Open a new tab and check if reputable people and organizations confirm what you’re seeing. 

In an era of dwindling trust, we should not fork over ours to cheap signals or cheap software.

Look closely at a painting, and you might be able to see where an artist changed their mind. Beneath the finished surface, there may be an abandoned line, a figure moved several inches to the left, a color that looked splendid in the morning and became unbearable by night. There may be a smudgy fingerprint at the edge of the canvas. The back may hold a date, a repair, uneven staples, a signature, or the name of a previous owner. These messy marks were once incidental to the artwork. Today they are considered evidence.

AI can produce an image in nano-seconds, then create hundreds of variations before a painter has mixed a color or cleaned a brush. Some of the results are technically bewildering and others inspire disbelief: How is this possible? Yet their abundance is changing our relationship to novelty. When nearly any visual idea can be summoned on demand, newness becomes easier to manufacture but harder to believe in. In an age of effortless images, the mess left behind by making something may become part of what makes it valuable.

So much recent visual culture can appear at once new and strangely familiar. Generative systems learn from immense archives of existing images, then assemble patterns according to statistical relationships. Their output can startle us, though the surprise often arrives cloaked in a recognizable visual fluency: the cinematic light, the dreamlike architecture, the glossy portrait or the eerily composed impossibility. After enough exposure, even the surreal can begin to feel conventional.

A study presented at the 2024 CHI Conference proffers one perspective about how this happens. Participants who used an AI image generator during a visual ideation exercise became more fixed on an initial example and produced fewer ideas, with less variety and originality, than those in the control group. The study involved only 60 people, so its findings should be treated cautiously, though they articulate something many artists and designers have begun to notice: a tool built to accelerate imagination can also narrow it by presenting a polished answer before a more difficult, original idea has had time to form.

Against this flood of frictionless imagery, fine art acquires another kind of presence. A physical artwork occupies time and holds emotion. An artist stood before it, considered it, changed it, ruined it, waited for it, and eventually decided to either finish or abandon it. All artwork contains a sequence of often messy choices that could have gone in any number of directions. The object becomes a record of attention, and attention is becoming one of the scarcest materials we have.

Researchers have already observed a version of this shift. Across six experiments involving 2,965 people, participants consistently valued art labeled “AI-made” less than work labeled “human-made,” even when the images themselves were identical. Awareness of the machine changed the perceived value of the human hand.

This preference extends beyond art. A 2015 study in the Journal of Marketing found that people considered handmade objects more attractive in part because they believed those objects symbolically contained the maker’s love. The participants were projecting on and responding to an imagined human presence embedded in the object, expressed through the perceived time and care required to make it.

Nostalgia is part of the current attraction to the handmade. We remember darkrooms, paste-up boards, paint-stained clothes, and the physical tools that digital technology has made unnecessary or irrelevant, and often forget the expense and tedium the processes demanded. Still, the renewed power of visible erasure or the pleasure of encountering an unrestrained line reaches beyond nostalgia. These marks allow us to envision the intent and the decisions behind them.

AI can attempt to imitate these marks. It can add a false thumbprint, invent pentimenti, reproduce a stray brush bristle, and generate an image that appears to have survived a century in an attic. But mess itself will soon prove very little. Provenance and an intelligible creative process will become more important as appearance ceases to represent authenticity.

In educated hands, AI can still become a profound artistic medium. Photography and digital tools have disrupted ideas about skill and authorship before becoming vehicles for extraordinary art. AI will be shaped by artists whose choices are specific enough to resist its defaults, and those artists may create work we cannot yet imagine. The presence of a tool has never settled the question of artistic value; the depth and consequence of the decisions made with it still matter.

We are entering a period in which polished images and endless variations are available on command. The works we retain or collect may be those that let us encounter another person’s limits: the place where control faltered or the artist had to surrender to something unresolved. 

I think our nostalgia for mess is growing from a hunger for evidence that making something still requires time, effort, commitment, and—dare I say it—love.

Everyone has a vision of the ultimate future, some picture of what the world looks like at its best (or worst), and whether that future is even reachable. In religious studies, we call this an “eschatology.” 

Eschatology isn’t just the Book of Revelation or the Rapture. The Greek meaning of eschaton is “last things,” and today the word captures the set of beliefs a person holds about where things are headed and what that implies for the present. 

You don’t have to be religious to have an eschatology. Political optimism is an eschatology, whatever side you’re on. Zohran Mamdani, a democratic socialist, thinks the future is something we build our way toward with free childcare and buses. Stephen Miller, one of the architects of the current immigration crackdown, thinks it’s something we had and lost, but can get back through sealing the border and mass deportations. Neither one just describes the future. Both are telling you what you’re obligated to do about it right now. In other words, your eschatology shapes your ethics, because what you believe is coming determines what you feel obligated to do today.

Silicon Valley has an eschatology too, and some of its leading figures have been pretty candid about what it is: AI surpasses humanity, humans become obsolete as a species, and what matters is making sure your version of AI is the one that wins. Understanding that vision as an eschatology, not just a business strategy, explains why some of the most powerful people alive seem so unbothered by what we seem to be losing—our attention spans, our critical thinking, our basic competence at things we used to just know how to do—and so uninterested in being accountable to anyone about it. 

Shades of Silicon Valley eschatology

To be sure, Silicon Valley CEOs each express their own flavor of this view of AI’s power.

Let’s start with OpenAI’s Sam Altman, because he was explicit about where he believes AI is going years before most people had even heard of a chatbot. In 2015, at a startup conference, he told the audience: “I think that AI will probably, most likely, sort of lead to the end of the world. But in the meantime, there will be great companies created with serious machine learning.” This doomsday eschatology was the basis for his business plan: race to the top because extinction is coming. 

Since then, Altman’s views have shifted widely. “Working towards prosperity for everyone, empowering all people, and advancing science and technology are moral obligations for me,” he wrote in an April 2026 blog post. “AI will be the most powerful tool for expanding human capability and potential that anyone has ever seen. Demand for this tool will be essentially uncapped, and people will do incredible things with it. The world deserves huge amounts of AI, and we must figure out how to make it happen.” 

This is a very different eschatology and a very different business pitch: we must invest in AI because it will help humans do powerful things. 

And then there’s the world’s richest man, Elon Musk, whose life’s work makes sense once you see the eschatology underneath it. The OpenAI co-founder has said that human extinction is a very real threat that he’s personally trying to prevent. “If you think about the future of humanity, it’s going to fundamentally bifurcate in two directions,” he said back in 2013. “Either it’s going to become multiplanetary, or it’s going to be confined to one planet, until some eventual extinction event.” 

You could argue that SpaceX exists as insurance against that second path, a hedge against asteroids, supervolcanoes, pandemics, or an AI that decides to take us out. In this case, his eschatology is that salvation is possible if we can colonize space, but only for those who can get there. This may also be why he has said empathy is “the fundamental weakness of Western civilization.” As bleak as I may personally find this perspective, it is in line with his eschatology. After all, it is illogical to care for others when most of humanity won’t survive.

Since Musk made his now-famous 2013 remarks, he has continued his claims about an AI apocalypse. And during his recent high-profile lawsuit against Altman, OpenAI, and its president, Greg Brockman, for altering OpenAI’s original non-profit structure, Musk referred to his ominous eschatology once more. 

“The worst-case situation is where it is a terminator situation,” said Musk in April, “where AI will kill us all.” 

“We are not going to talk much about extinction in this case,” interrupted Judge Yvonne Gonzalez Rogers. “They got it, that’s enough.” 

By May, an Oakland jury unanimously dismissed Musk’s case. But based on his worldview, you could understand why Musk maintains that OpenAI should remain not-for-profit. OpenAI’s lawyers had another theory: maybe Musk is jealous of OpenAI’s financial success? 

Entrepreneur Peter Thiel has arguably taken eschatology the furthest. He’s taken to giving lectures arguing that the Antichrist of the 21st century is a “Luddite who wants to stop all science.” And he gives examples: AI researcher Eliezer Yudkowsky and climate activist Greta Thunberg. In his framework, caution and regulation are the apocalyptic threat. This is perhaps why, when Pope Leo XIV used his first encyclical this year to call for international AI oversight, Thiel said the Pope was effectively a “Chinese communist agent,” restraining America so that China can beat us in the technology arms race. That seems like a strange thing to say about a pope, until you realize in Thiel’s version of the future anyone suggesting limits is getting in the way.

What our eschatologies excuse 

Three tech leaders, three different pictures of humanity’s ultimate destiny—resignation, survivalism, apocalypticism. But I would assert that they share the same consequences. For one, in each version, the future isn’t something we build together. It’s something they get to build on our behalf. 

And notice who is left out of each version. Altman’s future has room for “great companies” and “expanding human capability,” but arguably less concern for the people whose jobs AI might make irrelevant in the process. Musk’s future may have some room for wealthy people who can book a seat on a colony ship but less for the rest of us. Thiel’s future has room for innovation, but little for those who want to question what we are innovating for and towards. In every case, the very people who have a stake in the future are not part of building that vision of that future.

This also helps explain why some tech bros seem so disinterested in philanthropy or profit-sharing. If you genuinely believe the human era is ending and something else is taking its place, it stands to reason you have no need to feel obligated to the people around you. Why invest in public health, or housing, or building a robust civil society, if the relevant future doesn’t include most of the people who’d benefit? Why fund a school, or fix a subway, or pay taxes toward anything you won’t personally need in twenty years? This eschatology doesn’t just fail to produce ethics. It actively excuses their absence. Empathy stops looking like a virtue and starts looking like a waste of resources on people who, in this version of the future, don’t really count.

And this is where the theology does its real work: it doesn’t just predict the future, it shapes what futures are possible by dictating actions today. If humanity’s time is about to expire regardless of what anyone does, then choosing not to help others doesn’t feel like a moral failure. It feels like realism. That’s the result of every eschatology like these: it allows cruelty to feel like an accurate read of the situation rather than a choice.

What makes this eschatology dangerous is that it is held by the people with the most resources to act on it. Musk, Altman, and Thiel aren’t theologians speculating about the end times. They’re building the thing their theology tells them is coming, funding the politics that protects their ability to build it, and treating anyone who asks them to slow down as a luddite or Antichrist rather than a citizen with a legitimate concern. A belief that the present doesn’t matter, held by people with the power to shape it, doesn’t stay a belief. It becomes policy, product, and precedent, with tech bros deciding the future on behalf of everyone else, without anyone else’s consent.

Today, if you wish, you can go get a staggering amount of granular data about your health without your doctor’s involvement. All on your own, you can get an MRI of your entire body, a CT of your coronary calcium, an ultrasound of your blood vessels, and maybe someday an immersion ultrasound in a spa. 

Direct-to-consumer medical imaging tests are a huge and growing business worth billions of dollars. Over 100,000 Americans are already customers. But do these test results actually make us healthier? 

As a radiologist, a doctor specialized in the alchemy of turning information into health, I  translate data hidden in images into words that motivate behavior to promote healthier, longer lives. Because it is central to what we do, radiologists have grappled for decades with the question of whether and how information improves health. It turns out to be surprisingly complicated. 

In order to make a difference in someone’s health, the test must accurately predict or diagnose a condition, resulting in a change in mindset and behavior, and that change in mindset and behavior must produce a better outcome than not doing the test. 

For example, whole-body MRI of the general public, a test not recommended by any medical guideline, medical society, or payer, creates information by taking a picture of your body. That information is accurate for some diseases but not others. It might change your mind by finding something you didn’t know about, like a small mass in your kidney or thyroid gland, which then might change what you or your doctor does. For example, your doctor might order another test to better understand if the mass is important or perform surgery to take it out. 

This seems logical. But the reason whole-body MRI of the general public isn’t recommended while other screening tests are is because in the absence of evidence, it is believed to be unlikely to produce a better outcome than doing nothing at all.

Many people think, as I once did, that early diagnosis always improves health. That logic is especially intuitive for an early diagnosis of cancer. But in many cases, that intuition, however understandable, can be wrong. Finding something with a diagnostic test might not help, despite creating an illusion that it does.

For both examples I gave—small thyroid masses, small kidney masses—it has been well documented that aggressive efforts to treat them, despite many being cancer, does not help most people. Isn’t that strange? Yet despite ourselves, we keep finding and removing them.

That’s the tricky thing with diagnostic tests. It’s terrifically hard to unsee something. For those who receive no benefit from treatment, only secondary harm results: surgical pain and complications, the anxiety of a scary diagnosis, and financial burden. 

There are numerous non-intuitive examples like this in which only in retrospect do we realize patients have been harmed because a diagnostic test and its downstream actions were promoted to millions before confirming things worked as intended. This is why scientists and medical professionals are so obsessive about evidence and statistics. It’s really hard to know if something is hurting or helping if we rely on common sense.

The likelihood of benefit and the likelihood of harm matter tremendously. For example, if whole-body MRI is performed on a high-risk patient with a cancer-causing genetic syndrome, it is much more likely to create health benefits than if it were to be performed in the general population. This is one of the paradoxical things about diagnostic tests. It’s not just about test accuracy. Equally important is the people being tested, because baseline risk strongly influences the benefits.

Harm is often caused not by the test itself, but from actions taken to manage the information of unknown significance it produces. It is in this way that a seemingly harmless test can hurt people. We do tests to find things, and when we find things, we commonly act, even when we shouldn’t, because we aren’t rational decision-makers. 

Direct-to-consumer testing will change how medicine is practiced. It empowers us to make personal decisions about our health and generates novel, complex information that will unlock scientific breakthroughs. 

When buying one of these tests, we should pay attention to what the company is claiming. Is their test a curiosity, an entertainment product, a research tool, or clinical care? The lines are often intentionally blurry. 

The need for scientific support, informed disclosure, and regulatory compliance increases as the test moves from curiosity toward clinical care. If the test has only been shown to provide information or to make a diagnosis, or has only been shown to improve health in patients with specific risk factors different from our own, we should be skeptical. As they say, buyer beware.

Many companies encourage us to assume their test improves our health. In such cases, the information we are buying may help us, hurt us, or be a placebo and do nothing at all.

Imagine that someone you love starts turning down opportunities to attend social gatherings like going to the movies, picnics, and parties. Then this person stops responding consistently to your texts, avoids eye contact when you do get together, and seems to be losing their energy or spark. They have trouble keeping up with daily routines and talk a lot about physical struggles such as headaches and stomach aches. Something doesn’t feel right to you, but you’re not sure what is happening or what to do. Imagine now if the person all these things are happening to is you. 

Nearly one in seven people worldwide have a mental-health disorder. With the prevalence of mental disorders so high, odds are we all know and love someone who is struggling. But most of us have limited understanding of mental health, and sometimes misguided or inaccurate knowledge. If we improve our understanding, this could help us better identify when we are in distress and get the help we need. It can also potentially assist us in helping friends, family members, neighbors, and co-workers. As a psychologist, I believe we could all benefit from increased mental-health literacy.

The concept of mental-health literacy was first coined in the mid-1990s by a group of Australian researchers who began a series of studies to understand the public’s knowledge about mental health. Since then, extensive research on mental-health literacy has taken place in the U.S., Australia, Canada, the U.K., India, Japan, and elsewhere. 

Data from 16 studies indicate two major trends in public awareness about mental health. First, we’re making some headway in recognizing that mental disorders often have some biological underpinnings. Second, there seems to be more acceptance of professional help for such difficulties. But it’s not nearly enough.

Finding sound mental-health information can be challenging. The Internet has the potential to increase our access to evidence-informed knowledge and decrease unmet mental health needs, but there’s too much noise. And a lot of information online is of inadequate quality and lacks scientific evaluation. Rather than helping, random Internet searches on mental health could harm people most in need. 

One way to further increase our mental health literacy is to take the free online Mental Health First Aid course. This could help us gain knowledge on how to recognize when a disorder is developing and provide information on help-seeking options and available treatments. 

One of the best resources for evidence-based information is the National Institute of Mental Health. Established in 1949, this is one of the world’s largest research organizations funding brain and behavior research. It offers valuable online public resources including signs and symptoms of a broad range of disorders including anxiety, depression, eating disorders, and ADHD. It also provides information on the latest and best treatments and how to find a mental-health provider. Another accurate and up-to-date source is the National Alliance on Mental Illness, the largest grassroots mental-health organization in the U.S. They too offer research informed information online on how to find support including a helpline.

While mental-health literacy is everybody’s business, there are limits to the appropriate use of such information, particularly in our offers to help others. Let’s take the example of problematic drinking. It’s good to be able to recognize our own or a family member’s struggles with alcohol. But there are differences in providing someone with encouragement and assistance in getting help and respecting their boundaries if they refuse or are uninterested. Mental-health literacy can help us distinguish the best way to approach issues like these. 

If you or someone you know may be experiencing a mental-health crisis or contemplating suicide, call or text 988. In emergencies, call 911, or seek care from a local hospital or mental health provider.

It may be both the most prestigious and the most frustrating job in international diplomacy: members of the United Nations are weighing who should succeed António Guterres as Secretary-General when he completes his final term in December. The timing couldn’t be more critical, with U.N.’s fortunes at a historically low ebb, for reasons both political and financial.

Indeed, U.N. insiders question why anyone would even want the job at such a testing—and turbulent—moment. Some predict that, with major powers repeatedly blocking the U.N. from intervening in crises from Ukraine to Myanmar, and with the U.S. rolling back funding, whoever succeeds Guterres will simply have to continue to do “less with less” and manage a process of institutional decline as gracefully as possible. Others still hold out hope that a new U.N. leader could reverse the slide, and persuade powers large and small to invest in the organization.

Yet, at a time when global challenges—wars, epidemics, and climate change—continue to proliferate, there is a compelling case for preserving the U.N. as a space for global deal making. Doing so will require a Secretary-General of formidable political skills, capable of persuading world leaders to listen.

It is a question of balancing two competing imperatives. The U.N. is an organization that rests on international cooperation, but the current global moment is one of international competition. The challenge facing the next Secretary-General is to explain how the U.N. can stay relevant. This is no time to preach the virtues of solidarity among nations. It is necessary to meet governments “where they’re at,” as psychologists say, and accept their rifts are real. A shrewd leader of the U.N. could find ways to help states de-escalate conflicts and cooperate on global challenges out of necessity, not idealism.

Ultimately, if political leaders do not believe that the U.N., which was founded in 1945 to “save succeeding generations from the scourge of war,” can live up to that founding goal, they are less likely to take it seriously as an authority on other international issues such as Artificial Intelligence or economic cooperation. Making sure that doesn’t happen will be the central challenge for the organization’s next leader. Perhaps each nomination should thus come with a warning: Caveat emptor, or buyer beware. 

Our tumultuous present

Maybe the biggest challenge for the next Secretary General lies inside the cavernous chamber that houses the Security Council, the organization’s most powerful body. The five permanent, veto power-wielding members—the United States, China, Russia, Britain, and France—that dominate the council have repeatedly blocked intervention in crises from Ukraine to Myanmar. The Biden Administration’s repeated refusal to support calls for a ceasefire in Gaza at the U.N. in 2023 and 2024 stirred deep resentment across the organization.

Most member states still pay homage to the U.N. Charter and its prohibition against aggression. But Russia’s invasion of Ukraine and the U.S. decision to attack Iran in February suggest that some of the security council’s most powerful members no longer feel bound by its guiding principles. If the U.N.'s political situation is its main headache, its financial positions is a close second.

After the Trump Administration cut off the bulk of U.S. funding for its operations in 2025, U.N. officials have been spending a lot of their time wondering about how to pay the bills. The U.S. is supposed to cover about a fifth of the organization’s core costs and a quarter of the budget for its blue helmet peace operations, which add up to roughly $8.5 billion this year. Washington was already behind on its dues when Trump returned to office, and currently owes the U.N. about $4 billion. The U.S also previously covered roughly a third of the separate budgets for the organization’s big relief agencies—such as the World Food Programme—which need tens of billions of dollars to feed, vaccinate and shelter vulnerable people each year.

Other traditional donors, including European governments, are scaling back spending on development and humanitarian aid. Potential alternatives have not stepped in to close the financial gap. China is now on the hook for almost as much of the U.N.’s core and peacekeeping budgets as the U.S., but is often slow to stump up its annual contributions. Beijing also contributes very little to the humanitarian sector. While the Trump Administration has released some funds for the U.N.’s life-saving efforts this year, big relief agencies have been forced to cut back life-saving assistance to civilians in places like South Sudan and Afghanistan.

This is a markedly different landscape from the one António Guterres encountered in 2017, when the former Portuguese prime minister became Secretary-General. At that time, the U.N. members had just agreed to the Paris climate accord and a new set of Sustainable Development Goals, including a pledge to eradicate extreme poverty by 2030. Geopolitical storm clouds were gathering—the Security Council was fiercely divided over the war in Syria—and Donald Trump’s surprise victory in the 2016 elections, coming soon after Guterres’s selection, signaled trouble ahead. Still, for believers in multilateral processes, a broad agenda for international cooperation appeared to be in place, and the Secretary-General’s task was to turn it into reality.

That brief moment of optimism proved ephemeral. Guterres spent a large part of his first term trying, with some success, to keeping Trump at bay. Nonetheless rising tensions among major powers, and disputes over how to finance development projects and support poorer states’ climate change adaptation efforts, have eroded governments’ trust in one another, and their faith in the U.N.’s complex, negotiation-based approach to addressing their national interests. Guterres has spoken eloquently about the risks of international fragmentation, and has pushed U.N. members to work together on emerging challenges, such as the regulation of Artificial Intelligence. But he has struggled to get leaders to focus consistently on that agenda. Increasingly, many governments are turning instead to smaller, looser decision-making clubs—such as the Group of 7, a Western club, or the BRICS, the bloc of emerging market nations—rather than the U.N. itself.

Diplomats and U.N. officials also fault Guterres for taking a low-key role in crisis management, despite major conflicts exploding on his watch. The Secretary-General has been blunt about his lack of leverage to mediate in most of today’s conflicts, and has encouraged a culture of caution among his senior staff dealing with political affairs and peace operations. Although he was involved in one notable mediation success—the 2022 Black Sea Grain initiative—many U.N. members feel that his reluctance to engage in broader peace efforts has undermined the body’s credibility.

Runners and riders

Such critiques of the incumbent, and the manifold challenges facing the office of the Secretary General, haven’t deterred a clutch of candidates from around the world—who may yet be joined by new competitors as the race to succeed Guterres enters its final stretch. 

Eight candidates are currently jostling to replace him and assume the reins of what U.N. watchers sometimes call “the parliament of man.” Among the most widely discussed are Rafael Grossi, the Director-General of the International Atomic Energy Agency, and Rebeca Grynspan, a senior U.N. trade official and former vice president of Costa Rica. Carolyn Rodrigues Birkett, Guyana’s ambassador to the U.N., is also a popular candidate among her colleagues in New York. But because there is no official cut-off point for new candidates to enter the race, and many diplomats expect more challengers to emerge, especially if none of the existing crop looks like a clear winner.

While the current candidates have been speaking at public events from Europe to Korea, they have struck a cautious tone regarding their plans for the U.N. This reflects the way the selection process works. The Security Council—where the U.S., Russia, China, France and the U.K. hold vetoes—is central to the process, vetting candidates and recommending one to the General Assembly, where all U.N. members have seats, for approval. To date, the Assembly has always ratified the recommendation. Anyone who wants to be Secretary-General must avoid offending the five veto powers.

Talking about what is wrong with the world today without irritating at least one of the major powers is quite hard. Guterres angered Russia by criticizing its war in Ukraine, and Moscow is likely looking for a pliable U.N. chief. The biggest headache for Guterres’s successor will be maintaining working relations with Washington in the remaining years of the Trump presidency. Senior U.N. officials worry that, having imposed severe economic strain on the organization, the U.S. will insist that the next Secretary-General make deep cuts to U.N. agencies and follow the Trump Administration’s agenda on questions such as diversity, and soft-pedal criticism of Israeli actions against the Palestinians.

Tiptoeing around these sensitivities, the eight candidates for the Secretary-General position agree that the U.N. is in poor shape. All have promised to concentrate on preventive diplomacy, and have committed to managing the organization efficiently. None has so far set out a groundbreaking strategy for putting the U.N. back on its feet. If this is partly a matter of political tact, it also reflects the fact that the U.N. Secretary-General, despite being the public face of the organization, has only limited power to shape the organization. While the veto powers dominate the Security Council, a sprawling array of intergovernmental committees and boards—including dozens of entities dealing with issues from telecommunications to the environment—oversee the wider work of the U.N. system.

The Secretary-General and his or her advisers must barter with U.N. members over remarkably minor budgetary matters, down to the number of mid-level and junior staff employed in U.N. offices. It is easy to call on the U.N.’s top official to focus on problems of world order, but bureaucracy gets in the way. 

Once in office, the next U.N. chief will have to manage these institutional issues while also laying a broader agenda. She or he could address this challenge by appointing a powerful chief of staff to handle administrative matters, freeing up more time to prioritize diplomacy.

On the diplomatic front, the incoming leader’s priority will be to demonstrate the ability to insert himself or herself into one or more conflicts, signaling a shift from the Guterres era. This is easier said than done. Conflict prevention and peacemaking have become crowded fields, with a growing number of middle powers, and at times the Trump Administration, leading peace processes that the U.N. might once have overseen. 

The next Secretary-General will not simply be able to fly into trouble spots and cajole warring parties into making deals. But he or she can use the first months in office to shuttle between key capitals in regions such as the Middle East, looking for opportunities to engage, and encouraging U.N. staff—who still have considerable expertise on issues like mediation and peacekeeping operations—to table new options for dealing with ongoing or looming conflicts. This is not likely to produce quick or major wins, and there is always a risk that the U.N. will launch peace initiatives that fail. But as a European diplomat familiar with the U.N. once put it, everyone expects the U.N. to fail, and as a result it can do so without sustaining reputational damage. 

The greater danger for the institution right now is that it fails to engage in meaningful crisis diplomacy at all, cementing the impression that it is no longer a significant peacemaking force. If the Secretary-General can use this kind of early outreach to gain a place on the diplomatic map, she or he will still face the challenge of articulating what the U.N. stands for in a divided world.

Collective action in a world divided

There is no shortage of good arguments for why collective action helps address problems that transcend borders. These often get lost at a time when the predominant political narratives center on national divisions and differences.

U.N. officials do not help themselves by couching their arguments in technocratic terms, weighed down with talk of goals and policy frameworks. But the underlying political challenge is to make a case for multilateralism that resonates with decision-makers who view international affairs in zero-sum, or at least very low-trust, terms.

The best approach may be to frame the U.N. as a sort of global risk reduction organization. Over the last decade, governments, the private sector and the public have seen the knock-on effects of both the pandemic and regional conflicts causing global economic disruption. New technologies are creating further economic shocks and could make future conflicts more dangerous.

The U.N. is not a global authority with the powers to halt these trends. But it is a space in which governments can discuss the risks involved and—where they see common ground—explore steps to mitigate them. Even though it may be easier for states to meet in smaller clubs such as the BRICS and G7, the U.N.’s advantage is that it remains a neutral space where different blocs of governments can engage in dialogue.

The next Secretary-General can only encourage rivalrous states to take part in these dialogues, and will need to do so with a light touch. In the past, the U.N. has sometimes responded to new challenges by announcing new agencies and funds to address them. The resulting multilateral bloat has become hard to sustain financially and defend politically. The incoming U.N. leadership will need to show that it can facilitate discussions of difficult global problems without adding new layers of bureaucracy.

The role of the Secretary-General is perhaps to be both the world’s most prominent event organizer—bringing governments and other actors into the room to talk about their problems—and a sort of political therapist for world leaders. If the Secretary-General can win the trust of these leaders, and persuade them to work on common problems despite their divisions, he or she may be able to bring a degree of stability to a dangerous international environment. In doing so, she or he can also keep some political space open for the U.N. to do its most basic work, getting help to the suffering, mitigating the effects of suffering and – at its best – giving small states and civil society groups a rare platform to speak out on issues like human rights and climate change.

The new leader of the U.N. cannot heal all the world’s divisions, but with courage and political acumen, the Secretary-General can make international disputes a little less toxic. That may be a limited vision of what the U.N., sometimes dubbed the “parliament of man” is for.  But in an era of multiplying dangers and distrust, it is a noble and perhaps essential one.

Europe and Africa are at once impossibly close and worlds apart near the northern tip of Morocco. Ceuta and Melilla are two small cities overlooking the Mediterranean Sea and a short ferry ride across the Strait of Gibraltar from Spain. The cities on the African mainland have been controlled by Spain for centuries, and their residents are Spanish citizens who enjoy the privilege of European Union membership.

A border fence and a stark disparity in wealth and opportunity separate the two enclaves from Morocco: Ceuta lies a few miles from the adjoining town of Fnideq, and Melilla is 10-minute drive from Beninsar. In late July and early August, an estimated 72,000 people from Morocco—including many minors—walked, swam and ran toward Ceuta. Some made their way to Melilla, hoping the same rare fortune would await them there. Most carried nothing but a phone in a plastic holster and even lost their shoes to the waves, that later deposited them along the shore.

The people who made the hragga—colloquial for irregular migration—were driven by a combination of calculation and desperation. There were roughly three categories of people. Those who had tried before and had come prepared. There were those who had not been planning but had come to believe through a combination of online misinformation and unverified rumors, that a genuine window of opportunity to make it to Spain had opened and could close at any moment if they hesitated. And there were still others who embraced a daredevil approach with a combination of optimism and pessimism: “Why not? I have nothing to lose.” Many weighed the risks, reasoning that even a beating would be worth spending a day in Ceuta, which they are normally forbidden from entering.

Struggling to find food or shelter, most were either expelled by Spanish authorities or returned to Morocco voluntarily. Videos circulated online showing men kicked by the Spanish Guardia Civil while being expelled, then hit with batons by the Moroccan Gendarmerie on the other side of the border, in a choreography of rejection. The human cost was terrible. Some died falling from slippery rocks. Others drowned or were killed in the crush of thousands of bodies converging on the same narrow strip of shore. Among the 141 dead are: Faten, a smiling 20-year-old, talented footballer hoping to play in Europe; Ahmed, only 17; and many more. Parents of many youth who crossed over to Ceuta have been waiting in anguish at the border in Fnideq, carrying photographs of their missing children, checking with strangers if they happened to see their kids.

The timing was especially awkward for the Moroccan government, as the crisis erupted during the annual celebration marking the anniversary of King Mohammed VI’s reign. The government remained disengaged for days and attributed it to organized criminal smugglers. No one bought the argument, as it would be pointless to smuggle a person to the enclaves, since they are not part of the Schengen area. To reach Europe, a migrant would have to reach mainland Spain. Then, Rabat pointed to a Spanish Supreme Court ruling stating that migrants arriving by sea would not be deported automatically, as if legal nuance, rather than desperation in Morocco, was driving the migration.

The Moroccan government resisted acknowledging the reality behind this wave of migration: inequality, and the absence of work and opportunities for the country’s youth. In fact, those inside the country who spoke out against the state’s mishandling of affairs, and who reiterated why Moroccans wished to leave, were arrested. Only after growing pressure within Morocco did the government eventually agree to assist in identifying the unaccompanied minors and the deceased.

A profusion of conspiracy theories about geopolitics, circulating widely online, further obscured the actual forces driving the migration. Certain narratives, offered without any evidence, sought to portray the migrants as pawns in political games, disregarding entirely what drives and motivates people to leave home on such perilous journeys.

Repression and unemployment in Morocco

A constitutional monarchy, Morocco holds regular multiparty elections but the palace and its entourage dominate political and economic power through formal and informal means. Morocco has a repressive state, a deeply unequal society, and a profound crisis of unemployment and underemployment, especially for its young population.

From banning protests to fining editors to arresting rappers and  journalists, the crackdowns on political freedoms in Morocco have increased tremendously. While hundreds of the country’s brightest and most promising people perish in prisons, privileged Moroccans are more concerned with reputational damage to Morocco or insist that the country needs evolution and not a revolution. But how is evolution possible when a teenager demanding public hospitals instead of luxury football stadiums or unions pressing legitimate labor demands can get them beaten and arrested? 

Morocco’s economic model has produced soaring headline growth alongside deepening inequality, feeding a resentment that runs deep. Over the past two decades, capital has flowed toward megaprojects—the Tangier Med port, the Africa-Atlantic Corridor—financed by a mix of foreign and private investment. The services sector employs the largest number of people, followed by agriculture, forestry, and fishery.

The working age population of the country has grown by more than 10% over the past decade but employment has increased by only 1.5%, according to a 2025 World Bank report. The official unemployment rate stands at around 10% and at around 27% among young people aged 15-24. Yet 60-80% of workers in Morocco work in the informal economy, according to a 2025 report by the UN Economic Commission for Africa. The result is an economy whose glittering infrastructure sits uneasily alongside stagnant wages and rising prices for food, fuel and housing.

Unsurprisingly, many of the Moroccans who had expressed their frustrations and their desire for reform and opportunity in multiple waves of protests—the Arab Spring-era protests in 2011, the Hirak protests in 2016, the Gen Z protests in 2025—trudged by the thousands toward Spain. When loyalty flows only one way, the haves accumulate ever more while the rest are left to watch lavish banquets for Throne Day—and resignedly sigh, “hiwa hadchi li ‘ta Allah (This is the fate God has given us).” Hence, rather than asking, “Why do they leave?” we would do better to ask, “Why would they stay?”

Morocco has no shortage of resources: the country possesses rich agriculture and fisheries, phosphate, minerals, and enormous extractive projects in the occupied Western Sahara. A sense of being cheated is shared by millions as the economic benefits and opportunities do not trickle down to ordinary people. The contradiction between these natural resources and the everyday poverty, unemployment, and hardship experienced by ordinary Moroccans is expressed in the Moroccan chant that goes: “l'fosfat ou jouz bhorra / wa sha'bha mkhorra (Phosphate and two seas / yet its people still suffer).”

Moroccans see their government invest in the bling of glossy festivals and football stadiums built for the 2025 Africa Cup of Nations and the 2030 FIFA World Cup, which Morocco will co-host with Spain and Portugal. In the face of such visible wealth, it is what the state fails to do that carries the greatest weight.

A long transit in Tangier

Tangier is perched on the Strait of Gibraltar and has long served as a bridge between Africa and Europe. The city has drawn a rich tapestry of internal migrants from across Morocco, its diaspora, and the broader African continent.

In the early 1990s, when Spain—following a transformation of migration policies in the European Union and the implementation of the Schengen agreement—imposed more restrictive visa regimes, irregular migration began to accelerate. The International Organization for Migration estimated that irregular crossings to Europe through the Strait of Gibraltar rose from fewer than 5,000 in 1999 to around 20,000 in 2003.

Wars in Afghanistan and Iraq, military interventions and civil wars across Africa and intensifying climate catastrophes in the following two decades swelled global migration flows. Alongside African migrants awaiting passage, thousands from other parts of the world joined them. Europe, Australia and North America have all sought to obstruct migrants from coming in by any means necessary. The European Union outsourced border control offering economic and political incentives to Turkey and Morocco to stem the flow of migrants.

Under one EU-Africa agreement alone, $2.35 billion was allocated for deportations. This includes detaining migrants and dropping them in remote deserts. Several exit points along the so-called Eastern and Central Mediterranean routes were either sealed off by civil war in Libya, by special agreements with Turkey, and a bottleneck formed along the Western route through Morocco and Senegal.

Spain has already constructed a 1,600-foot floating barrier along the sea border following the Ceuta crossings. In the past decade, tens of thousands of migrants have died in Mediterranean crossings alone. But as Hein de Haas, a sociologist of migration, has explained, restrictive immigration policies fail to stop migration as there is a structural demand for cheap migrant labor in Europe. The waters of the Atlantic and the Mediterranean are full of the remains of those who perished while trying to reach Europe, victims of a system that treats human mobility as a threat to be managed rather than a reality to be humanely addressed.

But the combination of rising inequality and violence globally, and the need for a workforce, means people won’t stop migrating. This means that current border policies don’t stem migration but instead increase death and suffering. Precisely because it takes much longer, thousands of migrants are stuck in Morocco. Mehdi Alioua, a Moroccan sociologist, has called this phenomenon  “transit migration.”  Each must find rent and work, saving for yet another attempt at a better passage. Some have children, who grow up speaking fluent Darija, the Moroccan colloquial tongue, and who eventually become part of the working class fabric of Tangier. Some migrants with means share rented apartments in the affordable, peripheral neighborhoods of Mesnana and Boukhalef in southwestern Tangier.

Migrants without any financial resources, and mostly from other parts of Africa, camp in the woods in the forested outskirts of the city, where conditions are bare. When I visited their makeshift dwellings to meet some of the migrants, a strange odor hung in the air. Sticky puddles drew flies. I moved past the dumping ground and reached their improvised living quarters, which were kept meticulously clean.

During my fieldwork in Morocco between 2013 and 2018, it was obvious that many African migrants in long transit in Tangiers endured dire living conditions and faced racism. Black migrants were often called “Mamadou,” and sometimes even the dreadful “Ebola.”

In December 2013, Cédric, a young Congolese man, fell from his apartment rooftop in Boukhalef while being chased by the police. Many believed he had been pushed. Images that circulated on migrants’ phones showed his head smashed against the granite floor. Two days after his death, I saw the putrid pool of dried blood, nearly black, where he had fallen, when I visited his friends to pay my respects. What stayed with me for months was that I could still make out the pattern of the spatter from the impact.

Cédric’s death was a key turning point. A growing number of Moroccan activists understood the necessity of rejecting the notion that, in a poor and authoritarian country, organizing against racism is a “privileged” or “European preoccupation.” Moroccan activists, drawing on their own experience of repression, understood how short the distance can be between dehumanization, complacency, and abuse. They grew more attentive to the living conditions of migrants and began to challenge the normalization of racism more explicitly.

On one of those days, I met Najiba, a young activist, near Bab Al Qasba, one of the gates leading to Tangier’s medieval city. She and her comrades were showcasing their posters in hopes of garnering support from passersby for a demonstration the next afternoon. A placard read in beautiful Arabic calligraphy: “Kuluna Afariqa (We Are All African.)” Her group got some attention: some people paused for a few minutes to watch before moving on; others muttered some remarks too.

Najiba fell into conversation with a young man, around 16 years old. He had mocked the group’s slogan and pressed her on why they bothered organizing against racism. “Wasn’t it naïve,” he asked, “to defend the Africanos?” She told him that everyone had been a migrant once, and that Tangier itself had known great waves of migration from people of the northern Rif region.

She asked him where his own family was from. They shared Rif ancestry. “Ah, a Riffi, as well,” I heard her respond. “Did you know they used to say the Riyafa were taking people’s jobs and houses when our grandparents first came here?” She took him seriously and made him see historical patterns. He was intrigued and the teasing stopped. By the end of it, she remarked, “So we’ll see you here tomorrow, yes?” He made no promise. But he did wave goodbye, and his laugh, this time, was genuine.

The road from Ceuta

Many people located Ceuta and Melilla on a map for the first time during the recent frenzy. It would be no surprise if eyebrows were raised upon seeing just how far Spanish territory extends. These were sites of Portuguese and Spanish slave-trading routes and military garrisons in the 16th century.

The crossing to Ceuta was neither the first nor will it be the last of its kind. Because of its exceptional scale, Ceuta laid bare, much clearly, the contradictions of humanity in miniature, from kicking down someone to having the capacity to walk in someone else’s shoes. While a Dutch-Moroccan celebrity tweeted tactlessly, “Moroccan swims to Ceuta with an iPhone 16 to get an iPhone 17,” a woman took off her own shoes and handed them to a barefooted migrant.

While Italian, French, Spanish, Dutch, American and British politicians ratcheted up the headlines, numerous border staff stretched their discretion to the maximum to handle the situation with dignity. When fascists across Spain sowed panic and rallied to stop “the invasion,” local communities marched and blocked an anti-migrant demonstration. Ordinary people came out to provide emergency support; groups such as No Name Kitchen documented the treatment of migrants. Hope lies precisely in these different approaches.

I have been thinking repeatedly about Najiba, the activist organizing in defense of African migrants in Tangier. There were migrants from elsewhere in Africa in Ceuta, and they would have had much less protection or support once deported. Najiba’s example gave me hope that, even in the most difficult circumstances, solidarity can arise.

I think of the conversation between Najiba and the young man in Tangier: the boy’s smirk transforming into a thoughtful frown. I would like to think that something clicked in that moment. I wish I could reach back and speak with him. Standing there in Tangier, he was able to see things from another perspective. Perhaps he too instinctively recognized the moment for what it was. Or perhaps he, too, was among the thousands who had left for Ceuta.

The International Congress of Mathematicians met in Philadelphia last month, its first U.S. meeting since 1986. Minutes after accepting the Fields Medal, mathematics’s highest honor, Jacob Tsimerman told reporters he was leaving academia to work on AI safety. “I think the world is changing,” he said, adding that he did not expect the mathematical career to survive in its present form. 

Two months earlier, he had been one of nine mathematicians asked to review whether an AI model had truly refuted an 80-year-old central pillar of modern geometry, known as the Erdős unit distance conjecture. Tsimerman and his peers determined that it had, and their commentary alongside the 18-page proof helped translate the machine's argument into mathematics a human could understand.

The breakthrough, and subsequent report, reflected how the bottleneck that holds us back from forging new discoveries in mathematics, and, increasingly, in every other field, is changing. For most of history, the scarce resource was discovery. With AI, discovery is nonstop, and human confirmation is now what is scarce. 

To be clear, we are AI optimists. We believe AI tools will complement human ingenuity and expand what we can know and build, but our infrastructure for vetting and certifying discoveries was built for human throughput, and that is now the binding constraint. That is what holds innovation back. The answer to this problem is formalization: translating AI’s outputs into precise forms whose correctness can be checked automatically. Building the infrastructure to make verification routine is now a national-scale engineering problem.

Machines are outrunning us in more than math. The same is happening to the code that runs hospitals, banks, and power grids. In April, for example, Anthropic disclosed that its Mythos model could find unknown vulnerabilities in major operating systems and browsers, and restricted access to fifty organizations racing to patch them. Soon after, Microsoft engineers found 90 critical flaws in a widely used product; and in June, Sen. Mark Warner told a Senate hearing, citing the NSA director, that the tool "broke into almost all of our classified systems, not in weeks but in hours."

All of this arrives in a world that already runs on code we assume is correct. In July 2024, a single faulty software update, not a cybersecurity attack, grounded flights and disrupted hospitals worldwide. Nowadays, developers can “vibe code,” or use generative AI to write code with natural language prompts. This lets teams build quickly, but at the same time can produce vast quantities of code nobody fully understands.

The same AI systems accelerating discovery can also solve the resulting information overload — if we pair their generative power with the rigor of mathematics.

For decades, researchers in defense, intelligence, academia, and industry have built an alternative to blind trust: formal methods, systems for proving that software mathematically will do exactly what it was designed to do. Like the way a software compiler translates a program into instructions a computer can execute, formal verification takes code, a specification of what it should do, and a proof connecting the two, and accepts the program only if every logical step checks out. This makes it possible to automatically validate AI output, and more broadly rigorize software and other security guarantees. And generative AI tools have made this sort of formalization broadly accessible in a way it never was before. 

Once proof checking is cheap, the next bottleneck becomes writing the right specification. Deciding precisely what a system should do and capturing the world around it remains a matter of human judgment. A flawless proof of the wrong specification is still an answer to the wrong question. Formal verification cannot stop phishing, prevent every failure in a complex human system, or certify an AI model's overall safety—but it can eliminate whole classes of errors in critical software. For systems where failure can mean harm to millions, this should be a crucial step in the adoption of AI-generated code. 

The Leiden Declaration, endorsed in June by the International Mathematical Union, warns that AI threatens the verifiability of proof. We would put it the other way around: mathematics offers the infrastructure to make generative AI more trustworthy. For this reason, the United States should treat that mathematical rigor as a national mission. We need public infrastructure for verified software: open libraries of verified components and specifications, standards and benchmarks, better tools for checking updates, and training programs that connect mathematics, computer science, engineering, and national security—coordinated across agencies rather than confined to one program. And because formal methods rest on deep mathematical foundations, this requires sustained investment in mathematics research and education, not only in the applications built on top.

Companies operating critical infrastructure should not rely solely on reactive patching and bug bounties where failure could hurt the public. Instead, they should aim to build in auditing and formal guarantees from the start. The goal is not to formally verify every line of every program, but to identify the logic of critical systems, specify what that logic must and must not do, and achieve mathematical assurance in that logic.

Mathematics has weathered crises of trust before. Geometry, infinity, logic, and the nature of proof each threw the field into foundational doubt, and each time the response was more precise languages, stricter standards, and better verification. That rigor is what made mathematics scalable, and software now needs the same transformation. As AI writes more of the code the world runs on, blind trust is no longer good enough. The question is whether the United States will lead the shift to mathematical trust, or wait for the failures that force it.