Dr. Morgan Frank
AI’s Impact on the Early-Careers of College Graduates
Public debate often blames the difficult U.S. labor market of 2022–2023 on the rapid spread of generative AI, but systematic evidence on timing and distributional impacts is scarce. I combine monthly state unemployment insurance records with occupation and location data to measure unemployment risk across the U.S. workforce and show that risk in AI-exposed occupations began rising in early 2022, well before the launch of ChatGPT. Using millions of LinkedIn profiles, I find that recent college graduates suffered especially poor early-career outcomes, with gaps again emerging prior to late-2022 AI advances. Finally, drawing on millions of U.S. university course syllabi, I measure graduates’ exposure to large language model (LLM)–related content and show that greater exposure predicts higher starting salaries and shorter job searches after ChatGPT’s release. Together, these results demonstrate that labor market weakness preceded widespread LLM diffusion and that LLM-related education is associated with better—not worse—early labor market outcomes.
Morgan Frank is an Assistant Professor at the School of Computing and Information at the University of Pittsburgh. Dr. Frank is interested in the complexity of AI, the future of work, and the socio-economic consequences of technological change. Dr. Frank's research examines how individuals and skill-level processes around AI impact careers, firms, and society.