Peter McCrory
Anthropic researcher whose work measures AI's real impact on employment and hiring.
Last refreshed: 30 March 2026
Can an AI company's own researchers produce unbiased evidence of AI's effect on jobs?
Latest on Peter McCrory
- Who is Peter McCrory?
- Peter McCrory is an economist and researcher at Anthropic who co-authored a study measuring AI's real-world impact on employment using observed professional Claude usage data, rather than theoretical models.Source: Anthropic
- What is observed exposure in AI labour research?
- Observed exposure is a methodology developed by McCrory and Massenkoff that measures AI's employment impact using actual professional usage of Claude, rather than theoretical capability assessments. It found computer programmers face 75% task coverage.Source: Anthropic
- Has AI actually caused unemployment?
- McCrory and Massenkoff's Anthropic research found no systematic unemployment increase among heavily AI-exposed occupations since late 2022, but found suggestive evidence of slowing hiring among workers under 25 in exposed industries.Source: Anthropic
- How does Anthropic's AI job research compare to the Federal Reserve's findings?
- The Federal Reserve Bank of Dallas found employment down roughly 1% in the top AI-exposed industries, with the decline concentrated in workers under 25, largely through collapsed job-finding rates rather than layoffs. This broadly aligns with McCrory's finding of slowing youth hiring.Source: Federal Reserve Bank of Dallas
- Which jobs are most exposed to AI according to Anthropic research?
- Computer programmers face 75% task coverage by AI tools, computer and maths occupations 35.8%, and office and administrative roles 34.3%, according to McCrory and Massenkoff's observed-exposure methodology.Source: Anthropic
Background
Peter McCrory is an economist and researcher at Anthropic, specialising in the empirical measurement of AI's effects on labour markets. Working alongside Maxim Massenkoff, he developed "observed exposure": a methodology using real professional Claude usage to gauge actual AI displacement risk, rather than theoretical task-coverage models.
McCrory and Massenkoff's research found that computer programmers face 75% task coverage by AI tools, computer and maths occupations 35.8%, and office and administrative roles 34.3%. Workers in the most exposed roles are statistically older, more educated, and higher-paid. Their data showed no systematic unemployment increase among heavily exposed occupations since late 2022, though suggestive evidence points to slowing hiring of younger workers .
The research sits at the centre of a contested policy debate. The American Enterprise Institute cited similar findings when rebutting claims that AI worsens inequality, while the Federal Reserve Bank of Dallas independently found employment down roughly 1% in the most AI-exposed industries, concentrated among workers under 25 . McCrory's methodology is unusual: an AI company producing the most credible evidence about its own technology's labour-market effects.
