AI and Jobs: A Review

A review of theory, exposure measures, and evidence on generative AI’s labour-market effects — productivity gains are sizable but context-dependent, concentrated in high-wage occupations.

AI and jobs: theory, estimates, and evidence. We review research on generative AI’s labour-market effects across theory, occupational exposure measures, and empirical evidence from experiments and digital platforms. Productivity gains are sizable but context-dependent — 20-60% in controlled experiments, 15-30% in field settings — and effects concentrate in high-wage occupations. Evidence on employment is mixed: novice workers gain more on simple tasks, while platform data shows substitution in writing and translation work. We flag key gaps: most experiments test simple tasks on a narrow set of models, and adoption dynamics (whether exposure means substitution or productivity gains) remain understudied. Read more