AUTO-UPDATED

Dueling proofs from OpenAI and Anthropic astound scientists

OpenAI has released three significant mathematical and scientific proofs generated by its proprietary artificial intelligence models, sparking intense debate among experts regarding transparency and established research protocols.

Key Points

  • OpenAI claims to have solved the four-dimensional Kakeya conjecture across all dimensions, building on recent work by Fields Medalist Hong Wang.
  • The company announced a partial proof for the Yang-Mills existence problem, one of the Clay Mathematics Institute’s seven million-dollar Millennium Prize challenges.
  • A new proof refuting the Exponential Time Hypothesis (ETH) was released, though competitor Anthropic simultaneously published a conflicting independent result.
  • OpenAI’s release strategy contradicts guidelines from its own Advisory Group on Mathematics and Artificial Intelligence (AGMAI) regarding expert oversight and public disclosure.
  • The underlying prompts, computational data, and full methodologies for these proofs remain proprietary and inaccessible to the broader scientific community.

Why it Matters

These developments highlight a growing tension between rapid AI-driven discovery and the traditional peer-review standards required to validate complex scientific breakthroughs. The lack of transparency surrounding these proprietary models complicates the ability of the global research community to verify findings that could fundamentally alter fields like particle physics and computational complexity.
Scientific American Published by Joseph Howlett, Peter Hall
Read original