OpenAI announces three big science discoveries and a new way to work with academics

On Monday, October 26, OpenAI revealed three significant scientific breakthroughs across physics, mathematics, and computer science. The company disclosed that they had disproven the Exponential Time Hypothesis in computer science, a long-held belief in the field. In the realm of mathematics, OpenAI claimed to have solved two longstanding conjectures: the Kakeya conjecture and Stein's restriction conjecture in all dimensions.

The breakthrough in physics, however, was kept under wraps, described as a mathematically robust existence proof of specific interacting quantum field theories involving gauge fields and fermions in four dimensions, including Quantum Chromodynamics (QCD) and a unified extension of the Standard Model of particle physics. In response to previous criticism for releasing research findings without peer review, OpenAI adopted a new approach this time, involving academics as "shepherds" to guide the interpretation of their discoveries.

The company collaborated with external experts to provide guidance and structure the research, with recommendations varying based on the specific discovery. For the computer science and physics breakthroughs, external advice was sought, while the computer science research will be led by experts from UC San Diego and the University of California, supported by the company.