Anthropic‘s Economic Research team, in collaboration with independent researcher David Roodman, has released a review of evidence on worker retraining programs intended to mitigate labor disruption from AI. The review draws on 56 randomized US studies (combined in a new meta-analysis) plus experimental evidence from Europe. On average, job training programs produce modest but positive effects: for each person offered a training slot, employment rises by 2–3 percentage points and annual earnings by roughly $1,000, against a cost of about $13,000. After accounting for increased tax revenue and reduced benefit payments, the government recovers more than half of its spending, and programs roughly break even overall. A small subset of “sector programs”—which partner with employers in high-demand industries and place people directly into jobs—produce gains several times larger, but attempts to replicate these successes have often failed.

The authors conclude that if AI displaces workers at scale, existing retraining programs would likely fall short. Their central recommendation is to invest now in demonstrating, evaluating, and scaling the most promising approaches, including rapidly expanding a leading program for a specific group of workers and rigorously measuring results. The report is part of a broader Anthropic research agenda that includes an Economic Index tracking AI use across occupations, a framework for measuring AI’s labor market effects, and an Economic Policy Framework outlining possible responses. An Economic Futures Research Fund has been established to support investigations into such questions.