1 min read

Anthropic Research: AI Labor Market Impacts — New Measure Finds No Unemployment Rise Yet, Hiring Slowdown for Young Workers

Canopy Research

Anthropic researchers Maxim Massenkoff and Peter McCrory published a new empirical study introducing "observed exposure" — a measure combining LLM theoretical capability with real Claude usage data — to track AI's impact on labor markets. The study finds no statistically significant increase in unemployment for the most AI-exposed workers since ChatGPT's 2022 launch, but does find suggestive evidence of a 14% slowdown in new-job hiring for workers aged 22–25 entering high-exposure occupations. Computer programmers rank as the most exposed occupation at 75% coverage, followed by customer service representatives and data entry keyers, while 30% of workers show zero AI exposure.

Key Takeaways

  • "Observed exposure" metric combines Eloundou et al.'s task-level LLM feasibility scores (β) with real Claude usage data weighted by automated vs. augmentative use — computer programmers rank #1 at 75% task coverage
  • No measurable unemployment increase for top-quartile AI-exposed workers post-ChatGPT; but 22–25 year old workers show a ~14% drop in hiring into high-exposure occupations (barely statistically significant)
  • For every 10-point increase in observed exposure, BLS projects 0.6 pp lower 10-year employment growth — higher-exposed workers are disproportionately female, white, Asian, highly educated, and higher-paid

Original source: Anthropic Research