OpenAI claims to have solved a problem that defeated mathematicians for nearly 100 years, and the world's top mathematicians are not applauding.
In early September, a press release landed like a grenade in the quiet world of pure mathematics. OpenAI announced it had solved a major mathematical problem that had resisted human effort for almost a century. For most people, this sounded like unambiguous good news. For mathematicians, it felt like something stranger — and more unsettling.
What followed was not celebration but alarm. Twenty-five winners of the Fields Medal, which is widely regarded as the Nobel Prize of mathematics, signed an open letter warning of a "severe misalignment" between the goals of AI companies and the values of their discipline. That is not a small group of dissenters. Fields Medals are awarded only every four years, and winning one places a mathematician among the rarest minds on Earth.
When a quarter of all living medallists agree on something, it is worth paying attention. Colva Roney-Dougal, a professor of pure mathematics at the University of St Andrews in Scotland, has been watching this tension unfold with a careful eye. She argues that the way AI companies announce their breakthroughs — fast, loud, and aimed at headlines — sits uncomfortably with how mathematics actually works. In her field, a result is not a result until it has been checked, questioned, and understood.