NBER: AI writing in 29% of 2026 US STEM PhD dissertations

AI-generated writing appeared in 29% of US STEM PhD dissertations filed in 2026, up from 1.7% in 2023 and zero before then.

The finding comes from a National Bureau of Economic Research working paper by Daniel P. Gross of Duke, Dror Shvadron of Toronto and Hansen Zhang of Duke, covered by Nature. The authors ran the Pangram AI-detection service over abstracts from more than 226,000 US STEM doctoral graduates between January 2019 and May 2026. From the abstract: "AI writing was absent before 2023, appeared in 29% of 2026 dissertations, and is expanding quickly."

Among graduates from the same program and cohort, those whose dissertations contain AI writing are less inclined to pursue academic research careers and more likely to enter industry or non-tenure-track positions, with the effect "particularly pronounced for US-origin students." Nature puts the gap at 17% less likely to work in academia and 16% more likely to work in industry, for US-educated PhDs. In mathematical and computer sciences, 36% of abstracts were flagged. Among graduates educated in non-English-speaking countries, 41% of abstracts tripped the detector, against 23% from English-speaking countries.

"If our students are only doing half the work and they're dishing the rest to AI, do they really have the depth of understanding" for scientific careers, Gross asks.

The paper is a working paper and has not been peer-reviewed. The authors did not separate wholly AI-written text from partly AI-written text, though they report roughly 90% of flagged abstracts were entirely AI-generated. Two researchers in our Who's Who directory shared the link in the days after it appeared.