NBER: AI-written PhD abstracts tied to industry over academia

Nearly a third of US STEM PhD dissertations filed in 2026 contain text that an AI-detection service flags as machine-written, and the graduates who used it are routing to industry rather than the academic job market. The finding comes from NBER Working Paper 35859, with the write-up in Nature by Mohana Basu on October 16.

The authors, Daniel P. Gross of Duke, Dror Shvadron of Toronto, and Hansen Zhang of Duke, scanned more than 226,000 abstracts from US STEM dissertations completed between January 2019 and May 2026, running each through the Pangram classifier. "AI writing is absent before 2023, present in 29% of dissertations filed in 2026, and rapidly growing," they report. Math and computer sciences lead the fields at 36%. By country of origin, 41% of dissertations from students from non-English-speaking countries were flagged, against 23% from English-speaking ones.

The career-outcome result is the sharper part. Among US-origin graduates in the same program and cohort, those whose dissertations contain AI writing were 17% less likely to pursue academic employment and 16% more likely to work in industry, with the pattern "strongest among US-origin students."

Two sanity checks sit inside the methodology. Pangram flagged six of about 126,000 abstracts filed from 2019 through 2022, before public LLMs, implying a false-positive rate under 0.01%. The paper itself remains unreviewed.