OpenAI Releases AI Proof Challenging Computer Science Theory

OpenAI has published computer-science output from an artificial-intelligence model claiming to refute the Exponential Time Hypothesis, a long-standing assumption about the difficulty of solving certain problems. The proof remains under examination by outside researchers, and its validity has not been established. The Exponential Time Hypothesis, or ETH, holds that certain problems cannot be solved substantially faster than exponential time as their complexity grows. OpenAI’s result presents an algorithm for the 3SAT problem that runs slightly faster than exponential time, challenging the hypothesis. The company released the computer-science output publicly and asked outside experts to explain and assess it. Ryan Williams, a Massachusetts Institute of Technology computer scientist working with OpenAI on the review, said two or three parts of the proof struck him as unusual enough to warrant collaborative scrutiny. He is working with Russell Impagliazzo and Ramamohan Paturi, University of California, San Diego researchers who first proposed ETH, to examine the result. The group has also invited broader feedback on research practices around AI contributions. Anthropic announced...