AI-Exposed College Majors Face Weaker Early-Career Outcomes

Summary

A U.S. Census Bureau working paper examines how college graduates in majors with different levels of labor-market exposure to artificial intelligence have fared since large language models became available. Using administrative records, the authors find that employment, earnings, and job-switching patterns for the most AI-exposed majors began diverging immediately after ChatGPT was introduced in late 2022. In regression-adjusted estimates, graduates in the most exposed 10% of majors were five percentage points less likely to obtain initial employment, while their full-quarter initial earnings were 13% lower. The earnings loss was comparable in size to the loss associated with graduating during a major recession. About half of the earnings decline came from lower pay within the industry sectors employing these graduates; the remainder reflected movement into lower-wage sectors, including restaurants and retail. The effects became smaller as graduates moved farther beyond labor-market entry, but remained substantial for the most exposed majors.