8 in 10 young mathematicians have considered changing career plans over AI

The survey of 268 graduate students and postdoctoral researchers found that 56% of those who answered a question about career plans had considered changing them but remained undecided, while 17% decided against making changes and 8% decided to do so.

In a separate question, 75.5% said their likelihood of continuing in academic mathematics after completing their PhD or postdoctoral research had declined over the preceding months.

Another 76.3% believed AI advances would make it harder to secure a tenure-track position, a university teaching and research job that can lead to permanent employment.

The survey of 268 graduate students and postdoctoral researchers was conducted by Anna Brandenberger, Hanna Mularczyk and Dora Woodruff, three MIT doctoral students, and published on Oct. 29. Conducted between Sept. 17 and Oct. 8, it gathered responses from researchers at nearly 60 institutions across several countries. About 85% were graduate students, while the rest were postdoctoral researchers.

Graduating students from the MIT Department of Mathematics Class of 2025. Photo courtesy of MIT

The concerns come as AI systems become increasingly capable of solving complex mathematical problems. In early October, OpenAI released research manuscripts reporting progress across 372 categories of mathematical problems, including longstanding challenges.

The rapid advances have unsettled mathematicians, some of whom have described experiencing disbelief, shock, denial and anger before eventually accepting AI's growing role in their field, according to Business Insider.

For young researchers, the uncertainty goes beyond job prospects to whether years spent investigating difficult mathematical problems will remain valuable as AI improves. "It is difficult to do research and go to seminars when it feels like none of the individual work will matter much in the future," one anonymous survey respondent wrote.

Terence Tao, a Fields Medal-winning mathematician at the University of California, Los Angeles, has likened AI's rapid production of mathematical solutions to "strip mining" earlier research, according to the South China Morning Post.

Tao and 24 other Fields Medal winners warned in a joint letter that AI-generated mathematical solutions must be verified and integrated into the field by humans, arguing that shortcuts could undermine how mathematics is taught and understood.

Some survey respondents worried that AI could solve problems they had spent months investigating, while others feared that relying too heavily on the technology could weaken their mathematical intuition and independent problem-solving skills.

Despite these concerns, many were already using AI to explore academic literature, generate code and check mathematical arguments. However, 54.8% of those who answered a question about their experience using AI for research said they did not enjoy doing so.

The survey also revealed frustration over what some respondents considered insufficient guidance from academic institutions. "It'd be great if institutes gave guidance instead of platitudes on this subject," one participant wrote.

Others called for clearer policies from universities, academic journals and research funding bodies on the use of AI in mathematics.

The authors cautioned that participation was voluntary and the findings did not necessarily represent the views of the wider mathematics community. Some respondents also saw AI as an opportunity to accelerate discoveries rather than a threat to their careers.

The MIT researchers urged universities, academic hiring committees and senior mathematicians to take young researchers' concerns seriously. "Math needs motivated, supported graduate students and postdocs," they wrote.