HEIDELBERG, Germany — Mathematics is traditionally viewed as a slow, methodical discipline that rarely captures mainstream attention. However, the rapid ascent of artificial intelligence recently thrust top researchers into the spotlight at the Heidelberg Laureate Forum on September 15. The panel focused on how AI has begun to challenge the foundations of their field, having already contributed to solving dozens of problems posed by Paul Erdős, formally verified two Fields Medal-winning proofs, and advanced solutions for ten other longstanding open problems.
The controversy intensified on September 8 when OpenAI announced it had deployed 10,000 AI agents to tackle the Navier-Stokes existence-and-smoothness problem. This equation is one of the seven Millennium Prize Problems established by the Clay Mathematics Institute in 2000, carrying a $1 million reward for a verified solution. Only the Poincaré conjecture has been satisfactorily solved to date.
Tristan Buckmaster and Levent Alpöge, who had been working on the Navier-Stokes problem for approximately a year, alleged that OpenAI discovered their research and raced to beat them to the result. They further claimed that an OpenAI representative threatened them against releasing their own findings publicly.
Michael Harris, a Columbia University professor and host of the Silicon Reckoner newsletter, addressed the issue directly during the forum. Describing major tech firms developing AI as
Does anyone know if the Millennium Prize rules explicitly allow or forbid automated theorem provers? This feels like uncharted territory.
The threat to withhold publication sounds awful. Mathematics is about verification, not race, so speed shouldn’t trump ethical conduct here.
10,000 agents is wild scale. But I wonder if the Clay Institute has rules against AI submissions? That could make this moot.
Predatory tactics? If true, this is a disturbing precedent for human mathematicians competing against well-funded AI corporations.