Skip to content
wua. Poland
English

OpenAI’s Yang–Mills existence claim builds on Tadeusz Bałaban’s unfinished construction

Large Polish text sits beside a double chalkboard, with chalk lines on the left panel, a blank right panel, and two books on the desk below.
AI illustration · WUA

Two research teams used AI to extend a method Tadeusz Bałaban developed in the 1980s to construct Yang–Mills theory. In a PAP interview published on November 22, Wojciech Dybalski, a mathematical physicist at Adam Mickiewicz University in Poznań, explained what OpenAI claims to have achieved with that method: a proof that the theory exists. The claim does not cover the mass gap, the other part of the Millennium Prize problem.

Bałaban’s starting point was constructive quantum field theory. Researchers had developed its methods using simpler models with fewer dimensions of space and time. In a series of more than a dozen papers, Bałaban showed how to apply those methods to Yang–Mills theory, but he did not finish the construction. Dybalski recalled hearing specialists talk for years about taking up the work. Even piecing together the argument across the papers was a formidable task.

According to PAP, OpenAI researchers and, independently, Ethan Sussman and Levent Alpöge, who are affiliated with Anthropic, returned to those papers. Dybalski suggested that AI tools may have helped the teams both assemble Bałaban’s results and connect them with other methods. At the time of the interview, he said, OpenAI had released code in the formal language Lean but had not yet published a manuscript. Its proposed proof awaits peer review.

Physicists already use Yang–Mills theory to describe phenomena including the strong interaction. A rigorous construction, Dybalski said, could help them develop calculations with a known margin of error. The mass gap raises a different physical question: how massless gluons can bind together to form objects with mass. For Dybalski, there is good reason to recognise Bałaban’s contribution now, even though it is too soon to award the million-dollar prize for solving the whole problem.

Sources

Independent editorial content. WUA is not affiliated with any political party. 2026-11-22