New data show no AI jobs apocalypse–for now (www.brookings.edu)

A new report from Yale’s Budget Lab finds little evidence so far of a broad “AI jobs apocalypse” following ChatGPT’s November 2022 debut. Analyzing 33 months of occupational composition and unemployment data, researchers show the share of workers in jobs with high, medium, and low AI “exposure” (i.e., the percent of tasks where models like ChatGPT could save substantial time) has remained remarkably stable. While the study doesn’t rule out localized impacts—early-career workers and certain occupations may already feel effects—it finds no economy-wide displacement signal akin to a rapid collapse. The report also digs into adoption patterns and implications: generative AI use is highly uneven across sectors, with Anthropic’s Claude chats concentrated in coding and writing, and enterprise API use skewing toward automation (77% of business tasks) rather than augmentation. Occupational shifts since ChatGPT’s launch are only marginally faster than historical tech rollouts (computers, internet) and often predate ChatGPT, suggesting complementary investments, governance, and sector-specific frictions (privacy, liability, data access) slow diffusion. Yale will monitor impacts monthly and urges major AI firms to share granular usage data so policymakers and researchers can detect whether small, localized disruptions evolve into broader labor-market change.