OpenAI asegura que su IA resolvió uno de los Problemas matemáticos del Milenio

OpenAI announced on Tuesday that one of its artificial intelligence (AI) models has solved one of the Navier-Stokes problems, one of the six Millennium Problems that remain unsolved. According to the company, up to 10,000 AI agents collaborated for 88 hours to achieve the result. The discovery was made using an unpublished model that OpenAI claims is significantly more capable than their recently released GPT-6 Astra.

The Navier-Stokes equations, formulated in the 19th century, describe how fluids such as water and air move and are used in designing aircraft, predicting weather, and studying blood flow. The mathematical problem involves determining if a fluid that initially behaves regularly can develop a singularity in three dimensions, i.e., a point where its velocity grows without limit within a finite time.

OpenAI published a 165-page document supporting their claim that the equations cannot develop such a singularity within a finite time. However, the announcement has sparked controversy over the race to solve these problems and the potential access of AI models to unpublished research. OpenAI reportedly began testing their new model with the Millennium Problems on September 1st after noticing indications that two of these problems were nearing resolution.

They later explained that these indications were related to the work of Anthropic researcher Levent Alpoge and New York University mathematician Tristan Buckmaster, who had made progress on the forced Euler problem, related to fluid dynamics but distinct from the Navier-Stokes problem. Buckmaster stated that he and Alpoge used both Anthropic and OpenAI models to advance their research.

OpenAI denied accessing their work but acknowledged they cannot rule out that conversations with their AI systems contributed to the development of the tool used to solve the problem. The Millennium Problems were established in 2000 by the Clay Mathematics Institute, which offered a $1 million prize for the solution to each of the seven major mathematical problems.

To date, only one, the Poincaré conjecture, has been solved by Russian mathematician Grigori Perelman. OpenAI clarified in their announcement that they do not intend to claim the $1 million prize and that their aim in publishing the work is to demonstrate the advancement of their AI models.