What happened
Mathematicians from MIT, Harvard University, Northwestern University and New York University met on Monday at MIT to debate how AI agents are changing their discipline, MIT News reported. Science writer Kevin Hartnett moderated the session, which drew a full lecture hall and covered the quality of machine-made proofs, shifting research customs and career prospects in the field.
The backdrop was OpenAI's recent work. In September the company pointed thousands of AI agents at one problem and, after 88 hours, produced a proof connected to the Navier-Stokes equation, one of seven famed prize problems in mathematics. Panelist Tristan Buckmaster of New York University's Courant Institute and a collaborator at Anthropic were close to their own result when OpenAI announced. A month later OpenAI published 722 manuscripts claiming AI alone had solved many major open problems.
Harvard's Michael Hopkins called the papers nonsense filled with odd ideas, and Buckmaster said mathematics only matters if humans understand it. Northwestern's Bryna Kra argued the field should stop prizing being first and find new ways to measure understanding. MIT's Jacopo Borga questioned leaning on journals to judge mathematicians. Kra said a journal she manages faces a surge of AI-assisted papers and hired another managing editor.
Context
AI progress in mathematics built over a decade, often with help from mathematicians testing models on solved problems. In summer 2025, Google DeepMind models reached gold-medal level at the International Mathematical Olympiad, and other systems improved on some known solutions. Kra helped draft this year's Leiden Declaration on AI risks to mathematical research, and new venues such as the video journal Mathematical Discourse are emerging.
The briefing
The week in tech, in one email.
The stories worth your time, every week. Unsubscribe anytime.
