OpenAI's math AI model solves 372 major problems
OpenAI's new model found 372 solutions to major unsolved math problems.
OpenAI released less than 372 results from a new AI model. This model solves or makes major progress on big unsolved math problems.
The results include solutions to the four-dimensional Kakeya conjecture. They also show algorithm improvements and progress on the Riemann hypothesis, math's most famous unsolved problem.
The model created almost all results from a single prompt to one AI agent. Some results may have taken multiple tries.
This is much faster and cheaper than OpenAI's earlier work. That earlier breakthrough needed 10,000 agents and millions of dollars in computing costs.
Many results have been checked using Lean, a language that verifies proof logic. Mathematicians are skeptical.
They want to see exact prompts and compute times before accepting the claims. MIT's Andrew Sutherland represents this skepticism.
OpenAI has shared only average compute times. It has not shared individual details.
- 372
- Results released
- 1
- AI agents used this time
- 10,000
- Agents needed in earlier breakthrough
- Four-dimensional Kakeya conjecture
- Major conjecture solved
Why it mattersIf these results are proven true, they show that AI can solve major math problems faster and cheaper than people expected.
AustraliaNo direct Australian impact yet, but advances in AI problem-solving could influence research methods at Australian universities.
✓ Claims checked against the source and corrected before publish. checked 2 h ago
Open this story in InSnip →
