Frontier AI programs can now do better in math than graduate students.
It turns out that these students are not taking the news well.
In a paper published this week, MIT researchers found that a large majority of graduate students and postdoctoral researchers surveyed are seeing their career prospects dim rapidly as AI reshapes the field.
Nearly 76% said they were increasingly less likely to continue their math studies after graduating. 76% say they feel less motivated to do research “because of developments in AI.”
And in one part of the free-response survey, graduate students revealed their raw, pent-up feelings on the topic.
Perhaps most striking is the burning frustration with academic leadership, which respondents described as by turns capricious, unhelpful, meek, silent, and ineffective.
“It would be great if institutes gave advice rather than platitudes on this topic,” one respondent wrote.
“We lack high-level leadership within the university community,” said another.
For the survey, conducted by three mathematics students affiliated with MIT, researchers surveyed 268 early-career mathematicians via a Google form. The questions were multiple choice and free response. They all focused on AI’s ever-increasing ability to do work they’ve studied for years.
Notably, the response window closed just after OpenAI released hundreds of solutions to long-standing issues on October 6. The authors cautioned the survey, saying it represents the ambience of “only a small window of time.”
“The last few months have been exhausting,” the researchers wrote. “We are forced to ask ourselves existential questions about a field we love, while being incredibly uncertain about our own future.”
Since OpenAI burst onto the math scene in September, when it released solutions to the Navier-Stokes Millennium Prize question, AI-generated solutions have kept coming.
Mathematicians have described stages of disbelief, shock, denial, anger and, for some, acceptance.
Although some top math councils have come together to issue edicts on how AI should be used, it’s not like OpenAI should listen to them.
OpenAI and Anthropic have, however, made some concessions to the mathematical community. The two AI giants have pledged a total of $15 million to a foundation to support mathematicians, and OpenAI has created an “advisory group” to judge its mathematical functioning.
But if you ask students, the concessions haven’t changed much for them.
“It’s difficult to do research and attend seminars when it feels like no individual work will matter much in the future,” one interviewee wrote.
Gn bussni
