OpenAI announced three major discoveries in mathematics, computer science, and physics. The company added two findings to GitHub, including a proof for Sogge's local smoothing. Experts will guide the verification of these results. However, the AGMAI panel said AI companies should not give certain individuals first access to output.
OpenAI has announced three major discoveries in mathematics, computer science, and physics. The company had made a similar announcement less than a month ago when it released hundreds of manuscripts claiming to solve long-standing mathematical problems. This time, however, OpenAI chose to have external experts or shepherds guide the results in computer science and physics.
OpenAI has added two of its new discoveries to GitHub.
The first is a proof of Sogge's local smoothing for waves in every spatial dimension (at least two). It also yields the Kakeya maximal and dimension conjectures and Stein's spherical restriction estimates throughout the interior of the conjectured range.
The second discovery is a deterministic algorithm that counts satisfying assignments of fixed-width CNF formulas in subexponential time and polynomial space, refuting the Exponential Time Hypothesis.
OpenAI's physics discovery is a mathematically rigorous existence proof of certain interacting quantum field theories containing gage fields and fermions in four dimensions.
This includes Quantum Chromodynamics (QCD) and a unified completion of the Standard Model of particle physics.
Physicist David Gross will lead a group including Michael Douglas, Arthur Jaffe, Igor Klebanov, and Nikita Nekrasov to guide this work.
The computer science collaboration proves that the Exponential Time Hypothesis (ETH) is false.
OpenAI says verification of the result will be guided by a group of three experts including Russell Impagliazzo and Ramamohan Paturi, who were part of the group that originally proposed it in 1999, as well as MIT professor Ryan Williams.
The group has released an open letter to the community asking for input on how to examine this claim and set norms for dealing with AI-generated breakthroughs.
OpenAI's approach of giving certain individuals first access to its AI output has drawn criticism.
The AGMAI panel of math advisers recommended that AI companies should not follow this practice due to issues like unequal access, borrowed reputation, and biased selection.
Instead, they suggested future AI labs should release output on scholarly repositories with Lean-formalized proofs and information about how problems were chosen.