OpenAI is modifying its release strategy following advice from the scientific community. Less than a month after releasing hundreds of manuscripts containing solutions to math problems, the company has announced three new "major" discoveries spanning mathematics, computer science, and physics. The key difference this time is the involvement of external experts OpenAI describes as "shepherds" to guide results in computer science and physics, while for the math problem, advisers recommended releasing the AI output directly for researchers to develop their own explanations.
OpenAI has updated its GitHub repository with the latest findings. In the realm of computer science, the AI established a proof for the Local smoothing for the Euclidean wave equation, using an auxiliary partial differential equation to demonstrate Sogge’s local smoothing for waves in all spatial dimensions. It also successfully refuted the Exponential Time Hypothesis, providing a deterministic algorithm that counts satisfying assignments of fixed-width CNF formulas. An independent working group led by Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams will examine and explain this proof, releasing the materials immediately per the group's request.
The physics information has not yet been released in full, but it constitutes a "mathematically rigorous existence proof" of interacting quantum field theories containing gauge fields and fermions in four dimensions, including Quantum Chromodynamics (QCD) and a unified completion of the Standard Model. David Gross will lead a group including Michael Douglas, Arthur Jaffe, Igor Klebanov, and Nikita Nekrasov to shepherd this work. The mathematics breakthrough involves the model producing a proof of the local smoothing conjecture in every dimension, thereby resolving the Kakeya conjecture and Stein's restriction conjecture across all dimensions.
OpenAI acknowledges that this approach could present challenges. The AGMAI panel of math advisers recommended yesterday that AI companies do not adopt the practice of designating specific individuals to receive output first. They cited unequal access, borrowed reputation, and biased selection as significant concerns. Instead, the panel suggested that AI labs should release output on scholarly repositories, including Lean-formalized proofs, details on problem selection, and funding for researchers analyzing the results.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.