Economists and Nobel Laureates Warn AI Threatens Millions of Jobs

More than 200 economists, including 16 Nobel Prize winners, warned Monday that AI could reshape the economy faster than any technology in history. Governments, they said, have little time left to prepare. The four-sentence statement, titled “We Must Act Now,” was organized by Stanford University’s Digital Economy Lab and signed by economists, computer scientists and executives at Anthropic, Google and OpenAI.

The letter reads like rare consensus. Underneath it, the economists who organized it admit the profession still cannot agree on how much damage AI is doing to jobs today, let alone what happens next.

Four Sentences, 200 Signatures

The statement itself is short. It says AI “may become radically more powerful over the next 10 years” and that this “could drive an unprecedented transformation of our economy, larger than the Industrial Revolution, but unfolding over a vastly shorter time frame.” It warns of “large-scale job displacement” alongside “major gains in living standards,” and calls on leaders to “build the incentives, guardrails, and institutions needed to steer AI in a direction that complements humans and benefits society.”

Four economists organized it: Erik Brynjolfsson, who directs Stanford’s Digital Economy Lab; Ajay Agrawal; Anton Korinek; and Tom Cunningham. “We must act now to guide AI to complement humans rather than simply imitate them,” Brynjolfsson said in a statement released alongside the letter, adding that the goal is prosperity “for the many, not just the few.”

Michael Spence, the Nobel laureate and New York University professor emeritus, called for an “all hands on deck” response given the “scale, scope, and speed” of AI’s advance. Daron Acemoglu, the Massachusetts Institute of Technology economist and fellow Nobel laureate, said he was “happy to join other leading experts in calling for the urgent need to redirect AI so that its risks are minimized and it can work for the benefit of workers and society.”

The full signatory list runs into the hundreds and includes some familiar names:

• Yoshua Bengio, the University of Montreal computer scientist and AI pioneer, who said separately that AI’s trajectory makes it “highly plausible” that it will “drastically transform our economies”
• Jack Clark, a cofounder of Anthropic
• Eric Schmidt, the former Google chief executive
• Vinod Khosla, the venture capitalist
• the chief economists of both OpenAI and Anthropic

Bengio went further in his own statement, arguing market forces alone should not decide the outcome. “We must be intentional and make collective, democratic choices, rather than letting market forces play out and risking leaving most citizens behind,” he wrote.

Why Stanford’s Economists Chose This Moment

Economists spent years dismissing the loudest AI job fears as hype. ATMs didn’t replace bank tellers, the old argument went, so a chatbot wasn’t about to replace anyone else. That confidence has been cracking.

A month before the letter, Stanford’s Digital Economy Lab had already launched a dashboard tracking AI’s effects on jobs and growth in something close to real time, called the AI Economic Indicators. “This is not a one-and-done, static website,” said Christie Ko, the lab’s executive director. “It’ll be constantly growing and evolving to meet the challenges of the day.”

A separate working paper from the Federal Reserve Bank of Chicago, the Forecasting Research Institute, Yale, Stanford and the University of Pennsylvania surveyed 69 economists, 52 AI specialists and 38 superforecasters earlier this year and caught the same shift in real time. “Economists are certainly taking AI seriously,” said Ezra Karger, an economist at the Chicago Fed and one of the study’s authors.

The Early-Career Workers Already Showing Up in the Data

The clearest evidence so far sits in payroll data, not surveys. A Stanford analysis of ADP payroll records, the largest payroll processor in the country, found that employment for 22-to-25-year-olds in the most AI-exposed jobs, including software engineering, marketing and customer service, has fallen 16% relative to less-exposed peers since late 2022, even as overall employment kept growing.

The split runs by age as much as by occupation. Workers 30 and older in the same high-exposure fields saw employment grow 6% to 12% over the same stretch, the Stanford analysis found. Brynjolfsson, who co-authored the study, said older workers may hold tacit knowledge AI cannot easily copy, or simply more sway inside their companies.

A separate estimate, cited in the Chicago Fed led survey, put the same age group’s relative employment drop at 13%, not 16%. Two research efforts, both looking at 22-to-25-year-olds in AI-exposed work, landed on different numbers for the same basic trend.

Not every institution is shedding AI-related roles. The Pentagon has gone the other way, launching a hiring push aimed at hundreds of AI engineers to keep pace with military applications. The picture is uneven by sector as much as by age.

What Counts as “AI Exposure”?

“AI exposure” measures how much of a job’s daily tasks a language model could plausibly perform. Economists measure it five different ways, and those methods disagree most sharply on the very jobs everyone worries about most, including telemarketers, tax preparers and writers. No single number answers the question cleanly.

Torsten Slok, Apollo Global Management’s chief economist, laid out the problem in a blog post published about a week before the letter’s release. He compared five methods researchers use to estimate AI exposure, each producing a different answer.

The theory-based measures run higher than the usage-based ones, Slok wrote, because they ignore whether adoption is actually happening or worth the cost. Nela Richardson, ADP’s chief economist, has described much of the public debate over AI and jobs as “guesswork” given how many variables remain unresolved.

We are driving in the fog, and it is extraordinarily difficult to anticipate what will happen next. It’s the right time for a coordinated effort to bring clarity to a confusing situation.

Tom Cunningham, the METR researcher who helped organize the letter, said that in the statement Stanford released alongside it. Even Acemoglu, who spent years as the field’s most rigorous skeptic of AI productivity claims, once dismissed much of the broader debate as “brainless.” He signed anyway.

The Last Mass AI Letter Didn’t Stop Anything

This isn’t the first time the AI world tried the open letter approach. In March 2023, the Future of Life Institute published a call for AI labs to pause training of any system more powerful than GPT-4 for at least six months. It gathered more than 30,000 signatures, including Elon Musk, Steve Wozniak and Bengio himself.

No pause happened. Labs kept training larger models anyway. “We Must Act Now” asks for less than that letter did. It doesn’t call for a pause, a ban or a single named policy. It asks for “incentives, guardrails, and institutions,” language broad enough that almost any government response could claim to satisfy it.

That vagueness could be the letter’s biggest weakness: broad enough for almost anyone to sign, specific enough to bind no one.

Economists Bet on Retraining, Not Checks

The Chicago Fed led survey asked economists what policies they’d actually want, and the answers weren’t close. Targeted retraining programs drew 71.8% support. Universal basic income drew 37.4%. Government job guarantees drew just 13.7%. The general public, polled separately, was far more open to bigger structural fixes than the economists were.

The same researchers surveyed 69 economists and 52 AI specialists and found sharp agreement on one point: faster AI means fewer people working, paired with real economic growth. Under a rapid progress scenario, economists projected annual GDP growth reaching 3.5% by the late 2040s, while AI specialists in the same survey were more bullish still, forecasting 5.3%.

That growth would not spread evenly. The same researchers project the wealthiest 10% of households could hold 80% of total wealth by 2050 under a rapid AI scenario, a concentration higher than the United States saw before World War Two.

Some institutions aren’t waiting for Washington to move. Banks and other financial firms have already begun ordering compliance teams to rebuild their operating models before adding AI tools, rather than bolting AI onto workflows built for a different era. It is the kind of institutional retooling the Stanford letter says needs to happen faster, everywhere.

Frequently Asked Questions

Who organized the “We Must Act Now” statement?

Four economists organized it: Erik Brynjolfsson, who directs Stanford’s Digital Economy Lab; Ajay Agrawal of the University of Toronto’s Rotman School of Management; Anton Korinek, an economics professor at the University of Virginia; and Tom Cunningham, a researcher at the AI safety group METR. Stanford’s Digital Economy Lab published the statement on July 13, 2026.

What does the Stanford Digital Economy Lab track right now?

The lab runs a free platform called the AI Economic Indicators, launched in June 2026 with three dashboards. The Canaries Dashboard, built with ADP Research, tracks monthly employment by AI exposure level. The Takeoff Tracker evaluates evidence for explosive AI driven growth. The Adoption Monitor follows how fast workers and firms are actually using AI tools.

How many jobs do economists expect AI to eliminate?

Estimates vary widely and none are settled. The World Economic Forum’s Future of Jobs Report projects 92 million jobs displaced globally by 2030 alongside 170 million newly created, a net gain of 78 million positions. That aggregate hides which workers actually land the new roles, and how long the transition takes them.

Does the letter call for a specific policy like universal basic income?

No. The letter names no policy at all, only “incentives, guardrails, and institutions.” Separately surveyed economists lean toward targeted retraining over income guarantees, though the general public polled alongside them showed far more appetite for direct government intervention.

Which jobs are considered most exposed to AI right now?

Telemarketers, tax preparers and writers top the list under theory-based measures, according to Slok’s comparison, while usage-based data from real chatbot logs points more toward software engineering, marketing and customer service roles for young workers. Older workers in the same fields have so far kept their jobs.