OpenAI AI Model Disproves 1999 Computer Science Hypothesis

OpenAI announced that its artificial intelligence models have found a way to disprove the Exponential Time Hypothesis (ETH), a computer-science conjecture dating back to 1999, according to Decrypt. The announcement led to lighthearted confusion on social media among crypto traders mistaking the conjecture for Ethereum's ticker.

Disproving the Hypothesis

The Exponential Time Hypothesis was originally proposed by computer scientists Russell Impagliazzo, Ramamohan Paturi, and Francis Zane. The conjecture states that a classic logic puzzle called 3-SAT cannot be solved much faster than by trying every combination. Hundreds of subsequent research papers have relied on this assumption to demonstrate that problems in graph theory, computational biology, databases, and geometry cannot be solved quickly.

OpenAI published a 67-page paper detailing a deterministic algorithm that counts every satisfying assignment of such puzzles in subexponential time. According to the company, the AI found the disproof on October 16, and the result was formalized in Lean—software that mechanically verifies logical steps—on October 17.

Independent Examination

Original proposers Russell Impagliazzo and Ramamohan Paturi, alongside MIT's Ryan Williams, are leading an independent examination of the claim. In an open letter, the researchers noted that their involvement should not be taken as certification, emphasizing that a result of this significance requires careful, independent scrutiny.

Anthropic also released a paper credited to Claude on October 26 detailing a similar time algorithm for 3-SAT. Both companies are publishing data and contributing funds, with OpenAI allocating a $5 million contribution to the Simons Foundation to support conferences and programs examining AI-produced mathematical results.

Key facts

• OpenAI published a 67-page paper claiming its AI model disproved the Exponential Time Hypothesis.
• The ETH is a 1999 computer-science conjecture concerning the 3-SAT logic puzzle, unrelated to Ethereum.
• The AI discovered the disproof on October 16, with results formalized in Lean on October 17.
• Original proposers Russell Impagliazzo, Ramamohan Paturi, and MIT's Ryan Williams are independently examining the work.