Key facts
- OpenAI claims its AI has disproven the Exponential Time Hypothesis, a 1999 computer science conjecture.
- The hypothesis relates to the difficulty of solving the 3-SAT logic puzzle.
- OpenAI's AI reportedly found a subexponential time algorithm for counting satisfying assignments of 3-SAT puzzles.
- Mathematicians are independently verifying the claim, stating it requires careful scrutiny.
- Anthropic also published a paper claiming an AI model found a 2^o(n) time algorithm for 3-SAT.
OpenAI announced that its AI has disproven the Exponential Time Hypothesis (ETH), a significant 1999 computer science conjecture. The lab claims its AI developed algorithms that can solve certain complex logic puzzles, like the 3-SAT problem, in sub-exponential time, challenging long-held assumptions about computational difficulty. This development has sparked debate within the scientific community, with researchers like MIT Professor Ryan Williams leading efforts to independently verify the findings and discuss the implications for fields like cryptography. Anthropic has also released a paper with similar claims.