Particle.news
Download on the App Store

Anthropic Unveils AI Exposure Index, Putting Programmers at Highest Automation Risk

The new metric blends model capability with observed workplace use to reveal a wide adoption gap.

Overview

  • Anthropic’s “observed exposure” ranks computer programmers as most at risk, with about 75% of their daily tasks flagged as automatable in real-world use.
  • The top exposed roles also include customer service representatives, data entry keyers, medical record specialists, and market research analysts and marketing specialists.
  • Across Computer & Math jobs, large language models are theoretically capable of 94% of tasks, yet Claude usage currently covers about 33%, underscoring slow real-world adoption.
  • The report finds no systematic rise in unemployment in highly exposed fields since late 2022, though hiring of 22–25-year-olds in those occupations has fallen by roughly 14%, a result the authors say is only marginally significant.
  • Exposure skews toward older, female, highly educated, higher-paid workers, while about 30% of occupations—such as cooks, motorcycle mechanics, lifeguards, bartenders, and dishwashers—show near-zero coverage today.