Skip to main content

OpenAI's New Math AI Cracks 100+ Problems, Sparks Ethics Debate

OpenAI's Secret Math Model: Genius or Threat?

On September 22, OpenAI announced the formation of a Mathematics and AI Advisory Group, based at the Institute for Advanced Study in Princeton. The group consists of nine mathematicians who will operate independently from OpenAI. But that's not the headline. The real stunner? OpenAI revealed that a mysterious model, trained since August 28, has already solved over 100 long-standing open problems across multiple mathematical fields—including the notoriously difficult Navier-Stokes Millennium Prize Problem.

Mathematicians Shocked by Speed

OpenAI admits the model's progress caught even its own mathematicians off guard. "We were surprised," the company said, sparking internal discussions about how to break the news to the academic community. Mathematics, after all, is the bedrock of basic science. If AI can make new mathematical discoveries, the implications could ripple across every field. That's why OpenAI is treading carefully, studying how to deploy these capabilities responsibly.

But not everyone is cheering. Earlier, mathematicians Terence Tao, Yu Deng, Peter Scholze, and others published an open letter warning that using open math problems as benchmarks for AI could have negative consequences. OpenAI acknowledges these concerns, saying they highlight the need for deeper dialogue with the math community.

The Advisory Group: Independent, Unpaid, and Powerful

So who's on this advisory group? Nine heavy hitters, including three Fields Medal winners: Timothy Gowers, Martin Hairer, and Edward Witten. The full list: François Charles (ENS-PSL), Camillo De Lellis (IAS), Timothy Gowers (Collège de France, Cambridge), Martin Hairer (EPFL, Imperial), Nikhil Srivastava (Berkeley, Simons Institute), Ulrike Tillmann (Oxford, INI), Ravi Vakil (Stanford), Edward Witten (IAS), and Melanie Matchett Wood (Harvard). Notably, only De Lellis signed the earlier open letter.

The group operates independently from OpenAI. They can offer unsolicited advice, publicly comment on OpenAI's impact on mathematics, and even adjust their own membership. They don't get paid by OpenAI. However, they won't advise on the pace of OpenAI's internal research.

Their role? To help OpenAI assess the significance of new results, set academic standards for mathematical research, and recommend how OpenAI's tools can support math education and discovery.

What This Means for the Future

OpenAI's move is a clear acknowledgment that AI's foray into mathematics is a double-edged sword. On one hand, the potential for accelerating discovery is immense. On the other, the risk of disrupting academic norms—or worse, misusing powerful tools—is real. By bringing in independent experts, OpenAI is trying to build trust while keeping the reins on its own research.

But the big question remains: How will the mathematical community react to a machine that can solve problems they've wrestled with for decades? Only time will tell. For now, all eyes are on Princeton.

Key Points:

  • OpenAI forms an independent Mathematics and AI Advisory Group with nine mathematicians, including three Fields Medalists.
  • A mysterious model trained since August 28 solved over 100 open math problems in 24 days.
  • The group is unpaid, independent, and can publicly critique OpenAI's impact on mathematics.
  • OpenAI acknowledges concerns from mathematicians about AI's role in solving open problems.
  • The advisory group will guide OpenAI on result review, academic standards, and tool development for math research.