The AI Labor Apocalypse Is On Hold for Now

Summary

The article examines whether artificial intelligence is bringing the mass unemployment often predicted by the technology industry and governments. Its central conclusion is that a labor apocalypse has not arrived in the short term: employment across OECD countries remains historically high, while U.S. official statistics cited in the article show about one million AI-linked jobs created and roughly 200,000 destroyed over the past year. A Stanford study of job postings in 41 countries found that companies using AI most increased employment by 3.3%, although some weaknesses have appeared in consulting and programming. The International Monetary Fund estimates that AI will affect 60% of jobs in Western economies, and Anthropic has warned that Western unemployment could reach 10% or 20% by 2030; the article says such forecasts are not yet supported by broad labor-market data. Economists interviewed expect the first major effects to be qualitative, including reorganized tasks, fewer hours and possible four-day workweeks, rather than immediate mass layoffs. They nevertheless warn that recent graduates and workers in exposed occupations may face a narrower entry path, even if aggregate unemployment stays low. The article also emphasizes distributional risks: AI and robotization could reduce labor’s share of income, widen the gap between highly paid jobs and low-wage work, and concentrate economic power in a small number of companies. Experts disagree on the eventual scale and timing because AI adoption and policy choices remain uncertain. They call for reskilling, safety nets, incentives that prevent excessive automation, possible taxes on capital or data centers, and investment in lagging regions. The article presents regulation as a way to manage the transition rather than stop AI, while noting that proposals from major companies may also serve their interests, including limiting open-source alternatives.