AI maths advances unsettle top researchers

Recent AI announcements about solutions to previously unsolved maths problems are forcing leading mathematicians to rethink how their field will work.

Since April, AI companies have regularly published solutions to previously unsolved mathematical problems to show the capabilities of their latest models. OpenAI released solutions to 719 previously unsolved problems last week.

Hugo Duminil-Copin, winner of the 2022 Fields Medal, told colleagues at the French Mathematical Society that his previous research routes were now gone and that mathematicians were facing “a new day.”

Stephane Mallat, recipient of France’s 2025 CNRS Gold Medal, said the developments had caused an emotional shock but argued that mathematicians would still be needed. He said mathematics was more than calculation and remained a way for humans to think clearly.

The Simons Foundation, a major supporter of mathematics, announced a $15 million programme last week, co-funded by OpenAI and Anthropic, to support mathematicians working with AI. The programme will fund workshops to assess AI-produced results and set standards for AI use.

Francois Charles, a mathematician and department head at the Ecole Normale Superieure in Paris, said AI was raising legitimate concerns about how the maths community is organised, including at doctoral level, but said he was not pessimistic about the scientific side.