PhD students who write dissertation with AI less likely to stay in academia

Roughly 30% of PhD dissertations in science, technology, engineering and mathematics (STEM) in the United States this year contained AI-generated text, according to a working paper¹ that reviewed the abstracts submitted to an online database since 2019.

What’s more, the working paper, which has not been peer reviewed, found that PhD graduates who use artificial intelligence to help write their dissertations are more likely to work in industry and less likely to enter academia.

Various studies have shown a marked increase in the presence of text generated using AI in scientific literature over the past few years. Researchers at the US-based non-profit organization the National Bureau of Economic Research (NBER) wanted to understand how PhD students use AI because insights into students’ training and capabilities will shape science over the next several decades.

“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” to be good scientists? asks co-author Daniel Gross, an economist at NBER who is based in Durham.

The working paper analysed dissertations from more than 226,000 STEM doctoral graduates completed between January 2019 and May 2026 at US universities. The team used the AI-detection service Pangram to assess the abstracts. For the study, the researchers categorized abstracts as either AI-written or not; they did not distinguish between text that was wholly or partly generated by AI. (Although Gross says that 90% of the flagged AI-generated abstracts were classified by Pangram as entirely AI-generated.)

The team found that AI-generated text in dissertation abstracts first appeared in 2023, and rose from 1.7% that year to 18.8% in 2025. For dissertations filed during the first five months of 2026, 29.4% of abstracts contained AI-generated text. AI use was highest in mathematical and computer sciences: 36% of the abstracts had been written using AI. About 41% of dissertations by graduates educated in non-English-speaking countries had AI-generated abstracts, compared with 23% of dissertations by those from English-speaking countries.

Ang Liu, an engineer at the University of New South Wales in Sydney, Australia, says that one issue with PhD students using generative AI tools to write their dissertations is that researchers are expected to explore areas in which the answers are not already known. AI can be extremely helpful when students are working in areas in which there is already a substantial body of knowledge, he says. However, the challenge comes when research moves into areas in which knowledge is incomplete or uncertain. “That is where rigorous research training becomes even more important,” he says.

Job pathways

The researchers also reviewed PhD graduates’ education and employment profiles on the social-media platform LinkedIn. Among PhD graduates who completed their undergraduate degrees in the United States, those with AI-generated abstracts were 17% less likely to be employed in academia than were graduates who did not use AI. However, using AI made those graduates 16% more likely to be employed in industry than those who did not use it. The authors found that the link between AI use and landing a job in academia was less pronounced for graduates who did their undergraduate studies outside the US.

References