Thousands of artificial intelligence agents from OpenAI tackled one of the most complex mathematical problems in 88 hours, according to Phys.org, as reported by infohub.kz.
The developers launched an unannounced AI model in late August. The systems were combined into a network of 10,000 digital agents to solve extremely difficult mathematical problems. Over several days of continuous computation, the AI agents exchanged millions of messages and generated about 130 billion tokens, which would have cost roughly $10 million at standard rates.
The main achievement was finding a solution to the Navier-Stokes problem, which scientists have struggled with for about 90 years. This problem describes the behavior of moving fluids and gases and is included in the Millennium Prize Problems, for each of which the Clay Mathematics Institute promises a $1 million reward. OpenAI stated that it published the proof in the Lean system and does not claim the monetary prize.
The announcement has already sparked discussions in the scientific community. Independent mathematicians have not yet confirmed the correctness of the proof, and some researchers have mentioned a possible dispute over authorship and approaches to the solution. Nevertheless, experts call this experiment an important milestone in technology development, demonstrating AI's ability to tackle fundamental scientific challenges.