Over 200 economists, including Nobel laureates and AI company leaders, have jointly admitted the profession cannot clearly forecast AI's economic impact and is dangerously behind on understanding the question. They warn AI could drive transformation larger than the Industrial Revolution within a decade, risking large-scale job displacement, and call for policymakers and technologists to act now to build institutions and guardrails to steer AI beneficially. However, even basic measurement of AI's exposure to different jobs remains contested, with five competing frameworks producing sharply different conclusions about which occupations face the greatest risk.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Over 200 economists, including 16 Nobel laureates and the chief economists of OpenAI and Anthropic, signed a statement Monday titled "We Must Act Now" acknowledging that the profession lacks answers on AI's economic implications. The statement warns that AI may become radically more powerful over the next 10 years, potentially driving an unprecedented transformation larger than the Industrial Revolution but unfolding over a vastly shorter time frame.
Why it matters
The economists warn of risks including large-scale job displacement, though they also note opportunities such as major gains in living standards. The statement does not propose specific policies but instead demands that policymakers, economists, and technology leaders build the infrastructure—incentives, guardrails, and institutions—needed to understand and steer AI in a direction that complements humans and benefits society. Even skeptics like MIT's Daron Acemoglu, who has been rigorous in questioning AI productivity claims, have signed on, citing the need to minimize risks and protect workers.
What to watch
A key obstacle to clarity is disagreement over how to measure AI's impact on jobs. Apollo Global Management's chief economist identified five competing frameworks for measuring "AI exposure," ranging from tracking real worker usage of AI tools to theoretical expert assessments of job vulnerability. These frameworks disagree most sharply on jobs with the highest stakes—such as telemarketers, tax preparers, and writers—because theoretical measures run systematically higher than real-world usage data, ignoring whether adoption is actually happening or cost-effective.
The statement represents a rare moment of professional humility from a field usually confident in its analytical tools. Anton Korinek, the University of Virginia professor who organized the effort, frames it bluntly: "We are driving in the fog, and it is extraordinarily difficult to anticipate what will happen next." What lends weight to this admission is not just the number of signatories but their diversity—the group includes Daron Acemoglu, MIT's most prominent skeptic of AI productivity claims, and Erik Brynjolfsson, a Stanford economist known for bridging disagreement. Both have previously held conflicting views on AI's economic threat, yet both now agree the field lacks the conceptual and empirical toolkit to assess what is happening.
The deeper problem revealed by Torsten Slok's analysis is that economists cannot even agree on what to measure. The five competing frameworks for "AI exposure"—ranging from real worker adoption of tools like Claude or Copilot to theoretical expert judgment—produce systematically different conclusions about which jobs are at risk. The disagreement is worst precisely where stakes are highest: jobs like telemarketers, tax preparers, and writers show radically different risk profiles depending on which framework is applied. This suggests that much of the current public debate over AI and employment may rest on shaky conceptual ground, with different participants talking past each other because they are using incompatible definitions of exposure itself.
Brynjolfsson's own Canaries Dashboard offers one attempt to fill the gap, tracking 4.6 million workers across more than 730 occupations and showing employment shrinking more than 4% annually for workers ages 22 to 25 in AI-exposed occupations, even as headline labor market data appear calm. This detail underscores why the statement calls for urgent action: the danger may be invisible at the aggregate level and only visible once data are disaggregated by age and task exposure, a possibility that conventional economic monitoring would miss.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
