OpenAI forms math advisory board as its AI solves over 100 open problems

OpenAI is reportedly close to solving a second Millennium Prize math problem, even as its first such claim — a solution to the Navier-Stokes existence and smoothness problem announced on September 8 — faces mounting criticism from mathematicians who say the company exploited a loophole in the problem's formulation.

A Contested Victory and a New Frontier

The company's Navier-Stokes proof, produced by 10,000 autonomous AI agents running on an unreleased internal model over 88 hours, was hailed as a landmark when it was announced. But according to Scientific American, the proof solved a variant of the problem that relies on a contrived external force — an approach many experts consider disconnected from the physical reality the equations are meant to describe. "The most important problem is unsolved," Luis Silvestre, a mathematician at the University of Chicago, told Scientific American. "The Clay Problem is settled, but the main problem for the Navier-Stokes equations is not".scientificamerican+2

Three mathematicians have since posted a proof showing that OpenAI's method can never be extended to solve the full problem without the external force, meaning the approach has reached a dead end. The result technically satisfies the Clay Mathematics Institute's original problem statement, which included an option allowing external forces, but fluid dynamicists now face the uncomfortable possibility that the equations can only blow up under artificial conditions.scientificamerican

Meanwhile, The Information reported on Monday that OpenAI is close to solving another of the remaining Millennium Prize Problems, and OpenAI itself disclosed on September 21 that its new model "has now resolved more than 100 long-standing open problems across most areas of mathematics".openai+1

Mathematicians Push Back

The rapid progress has alarmed the mathematical community. On September 11, 25 Fields Medal winners published a joint declaration titled "A Severe Misalignment of AI in Mathematics," warning that AI companies' race to solve famous problems risks undermining the discipline's culture of open inquiry and human understanding. "Often these solutions are announced in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work of others," the letter stated. Separate accusations emerged that OpenAI pressured a mathematician not to credit a collaborator affiliated with Anthropic.techcrunch+2

An Advisory Group as a Bridge

In response, OpenAI announced on Monday an independent Advisory Group on Mathematics and Artificial Intelligence, hosted at the Institute for Advanced Study. The group's nine members include Fields Medal winners Timothy Gowers, Martin Hairer, and Edward Witten, alongside mathematicians from Harvard, Stanford, Oxford, and Berkeley. The group will advise on the review and communication of results, uphold academic standards, and operate independently — members will not be paid by OpenAI and may publicly challenge the company's approach.openai

"There are difficult questions ahead about how AI can support mathematical understanding," OpenAI wrote, adding that it wants "mathematicians to be at the center of shaping the answers". The group will not, however, advise on how OpenAI paces its internal progress — a distinction that may do little to ease the concerns of mathematicians watching the machines accelerate past them.openai