Labor market impacts of AI: A new measure and early evidence

5 hours ago

Introduction of a new measure called 'observed exposure' to assess AI displacement risk by combining theoretical LLM capability and real-world usage data.
AI's actual coverage remains a fraction of its theoretical capability, indicating significant untapped potential.
Occupations with higher observed exposure are projected by the BLS to grow less through 2034.
Workers in highly exposed professions tend to be older, female, more educated, and higher-paid.
No systematic increase in unemployment for highly exposed workers since late 2022, but hiring of younger workers in exposed occupations may have slowed.
Comparison of theoretical AI capability (β) vs. actual usage (observed exposure) shows AI is far from reaching its full potential.
Top exposed occupations include Computer Programmers, Customer Service Representatives, and Data Entry Keyers.
Workers in the most exposed occupations earn 47% more on average and have higher education levels compared to unexposed workers.
Initial analysis shows no significant impact on unemployment rates for highly exposed workers, but tentative evidence of slowed hiring for young workers (22-25 years old).
The framework aims to track AI's labor market impacts over time, with potential for future updates and refinements.

Hasty Briefsbeta