Sam Altman, the CEO of OpenAI, says mathematicians have already moved from fear to acceptance of AI. Speaking at a recent talk at Stanford, he described a five-day arc: shock, then anger, then a willingness to work with the technology. Not everyone in math sees it that way.
Altman's version
Altman said the first reaction was "We're cooked, and our field is over." He said some mathematicians even talked about telling grad students to switch majors. By day three, he said, the mood was anger that OpenAI had done this. By day five, he said, many were saying they would have a different but hopefully better career.
What the math world says
The backdrop is a rough few months. In September, OpenAI announced its models had solved the Navier-Stokes equations, a famously hard math problem that had stumped top minds since the 19th century. Prominent mathematicians were furious. The pushback came in several forms:
- Two academics, one at NYU and one at Anthropic, accused OpenAI of using their chat logs to help crack the problem.
- Mathematicians said the AI did not show its work, leaving them the job of checking the answer.
- More than 700 academics signed an open letter against a Caltech student contest, calling it a promotion of AI "slop."
The contest, called a "mathathon," encouraged participants to use AI on long-standing problems. OpenAI dropped its sponsorship. Organizers have since adjusted the rules, and the event is set for November.
A changing job
Terence Tao, a UCLA mathematician who earned his doctorate at 20, said math's "romantic era" is over. Top academics like him are shifting to a workflow of verifying AI's work instead of solving problems on their own. After OpenAI said this month that it had solved more than 700 additional problems, Tao said, "Some things are being lost."
Why it matters beyond math
Altman said math went through this cycle faster than any other field he has watched. If so, it offers an early look at what AI may do to expert work in general: less time producing answers, more time checking them. Altman also said he does not expect a "job apocalypse" as companies restructure their workforces.
What is not known yet
Altman's account comes from one talk. The material gives no survey of how many mathematicians actually feel accepting. The critics' concerns about chat log data and unchecked proofs are accusations and complaints, and the article does not report how OpenAI answered them.
The takeawayIn math, AI is shifting experts from solving problems to verifying machine answers, and the field is still split on whether that is progress or loss.