OpenAI Claims AI Refuted Exponential Time Hypothesis, Sparking Scientific Debate

OpenAI releases solutions to three major math problems, including a surprising rejection of the Exponential Time Hypothesis (ETH), a longstanding assumption in computer science about what is difficult to compute. As with previous proof drops from the lab, the scientific community responds with a mix of thrill and dismay. Anthropic mathematician Levent Alpöge suggests OpenAI bases its solutions on pre-existing work, and hours later Anthropic releases its own refutation of ETH.

OpenAI evolves its release process this time, tailoring it to each discipline: mathematicians receive the work on GitHub, while computer scientists form a working group requesting public proofs and a funded conference. MIT professor Ryan Williams, who previously called an earlier OpenAI proof dump a "DDOS attack on mathematicians," agrees to help interpret the solution alongside ETH's original 1999 authors, Russell Impagliazzo and Ramamohan Paturi of UC San Diego. Their first step is releasing an open letter seeking feedback and discussion of norms around AI contributions to science.

Williams notes that while some parts of the proof seem routine, two or three "alien-looking" parts convince him the whole community needs to work on it together. The ETH belongs to the broader class of P versus NP problems, which ask whether any problem whose solution is easily verifiable can also be easily solved. The debate underscores growing tensions over how frontier AI labs should handle scientific breakthroughs of this magnitude.

Read More at the original source →