OpenAI said to be nearing proof of Hodge conjecture, another top math prize
OpenAI is reportedly close to solving the Hodge conjecture, a major unsolved problem in mathematics, according to a person familiar with the matter. The company previously claimed a solution to the Navier-Stokes problem, though that result remains unconfirmed. OpenAI may delay any announcement to avoid a repeat of the public relations issues surrounding its earlier math breakthrough.
The Hodge conjecture, one of seven Millennium Prize Problems, asks whether certain geometric shapes can always be broken down into simpler algebraic components. OpenAI's reported work follows its earlier, still-unverified claim on the Navier-Stokes equations, which triggered a public relations backlash. The company is said to be using a pretrained model variant codenamed "Doug," with computing costs potentially reaching millions of dollars.
Internally, some researchers believe such mathematical breakthroughs could contribute to recursive self-improvement—AI systems optimizing their own architecture. However, the mathematics community has reacted with skepticism and even anger, with some practitioners viewing the developments as a threat to their field's traditional methods. OpenAI may postpone any formal announcement to better manage communications this time, learning from the previous controversy.
If OpenAI confirms a Hodge conjecture solution, it could reshape how mathematical research is conducted, potentially displacing human mathematicians in certain domains. Society may see accelerated scientific discovery, but also increased distrust if claims remain unverified. The broader public could face a widening gap between AI capabilities and human understanding, affecting education and funding priorities. The impact may be most acute for academic researchers, who could find their expertise devalued, while industries relying on pure math might gain new tools. However, the unconfirmed nature of the earlier Navier-Stokes result suggests caution is warranted.