# A mathematician cries over AI's math wins > Steven Strogatz on the week AI cracked problems his field spent decades loving, and what that costs the people who loved them. - Title: A mathematician cries over AI's math wins · WIRED | Shortify - Summary: Steven Strogatz on the week AI cracked problems his field spent decades loving, and what that costs the people who loved them. AI is answering the questions… - Keywords: mathematics, openai, anthropic, research careers, mathematician, cries, AI's, math, wins, WIRED - Source: WIRED — https://wired.com/story/mathematician-steven-strogatz-grapples-with-ai-recent-breakthroughs - Author: Isabella Ward - Published: 2026-09-12T10:00:00+00:00 - Read time: 2 min - Topics: ai, mathematics, openai, anthropic, research careers ## The tears came first Cornell professor Steven Strogatz starts to cry describing a single week of AI breakthroughs in mathematics. The science thrills him. The behavior around it does not. > The science is thrilling, but there's a lot of human unpleasantness going with it. ## A 90-year-old problem falls OpenAI said it threw tens of thousands of agents at a math problem carrying a $1 million prize, and solved it. The result still needs independent verification. ## A credit fight breaks out Mathematician Tristan Buckmaster says OpenAI rushed the problem after learning of his work with Anthropic researcher Levent Alpöge, and tried to shape who got credit. ## Labs racing toward IPOs Strogatz reads the rupture as a symptom: corporate labs chasing headlines ahead of enormous public offerings, with mathematics as the scoreboard. ## Claude proves 29,500 theorems Anthropic said Claude formalized an existing proof of Fermat's Last Theorem by proving 29,500 small theorems, work other mathematicians had been grinding at for years. ## No longer at the frontier Alex Townsend used ChatGPT to crack a decades-old linear algebra problem. He is also 15 years into a career whose peak skill just got passed. > I'm the one with an AI agent, which feels very different actually. And I feel totally threatened by it. ## Marketing, not engineering Strogatz is blunt about Navier-Stokes existence and smoothness: arcane, useless to a working engineer, and valuable mainly as proof one lab beat the other. > Nobody cares about the Navier-Stokes singularity problem; only a tiny subset of pure mathematicians care about that. ## Who should get the million He wants Córdoba and Martínez-Zoroa considered, and makes the case for Buckmaster, who posted solutions to three closely related problems days before OpenAI. > They were on the trail, and I think they would've gotten there, but we don't know. ## Humans as translators Machines produce proofs nobody can read. For now, explaining them back in human terms is a job only experts do well. Strogatz expects that to fall too. > Are we going to be the great interpreters of the things that the machines do? ## Mess is the moat The closer a field sits to the real world, the more it resists AI. Economics, sociology, international relations: safe for a while, because they are messy. ## Who has taste Infinitely many things are provable, but only some are interesting. Machines have shown no sign of mathematical taste yet, and no obvious reason they cannot learn it. ## Devastated or democratized The questions mathematicians love are getting answered. That is a loss if you loved the struggle, and an opening if you only ever wanted in the door. > Pure math is being devastated or revolutionized, depending on your point of view. ## The canary in the coal mine We still play tennis and chess badly and happily. But nobody funds a beautiful game, and math may be the first field to find out what losing understanding feels like. > I feel like we're the first battleground here in math where we're on the brink of losing human understanding. ## Key takeaway AI is answering the questions mathematicians built their lives around, and the open question is what human mathematicians are for once the answers arrive without them.