Nobel laureate validates OpenAI's AI-generated Yang-Mills proof, but $1 million problem remains open

On October 28, OpenAI submitted an existence proof tackling one of modern physics' toughest questions, the Yang-Mills problem. Now, a group of leading physicists has backed the existence result, although a crucial part of the famous problem remains unsolved. For those unfamiliar with it, the proof is a formal mathematical construction showing that a version of quantum Yang-Mills theory can exist, a framework linked to the strong force that holds the building blocks of matter together. Notably, AI companies are increasingly shaking up the scientific world with one breakthrough after another, with OpenAI's earlier work on the Navier-Stokes equations being another example.The assessment was published on November 10 by the Yang-Mills Working Group, chaired by David Gross, a 2004 Nobel Prize winner in Physics. The group examined a Lean-coded submission sent by OpenAI and made the full submitted code and its validation report public. OpenAI says its models produced a written proof on September 9 and completed the Lean formalisation on October 16. The assessment group also includes Michael Douglas, Arthur Jaffe, Igor Klebanov and Nikita Nekrasov.What is Yang-Mills theory, and why does it matter?Think of the strong force as nature's glue. It holds quarks together inside protons and neutrons. In quantum chromodynamics (QCD), the theory describing this force, gluons act as messengers carrying the force between quarks. Pure Yang-Mills theory strips away the quarks and focuses on how gluons interact with one another. It is a key mathematical building block for understanding the strong interaction.The challenge is not simply to use these equations to make predictions. Physicists have relied on Yang-Mills theory for decades, but a rigorous mathematical proof that the quantum theory exists and has the expected properties has remained elusive. The Clay Mathematics Institute made this one of its seven Millennium Prize Problems in 2000, attaching a $1 million prize to a complete solution. The reviewers say OpenAI's submission constructs a relativistic quantum field theory from four-dimensional lattice calculations. The work was formalised in Lean, a proof-assistant language that allows computers to check the logical steps in a mathematical argument. The group said its clean rebuild, axiom audit and other formal checks passed.However, the result does not solve the full Millennium Prize problem. The assessment says the submission "does not prove" the existence of a mass gap. In simple terms, a mass gap means there is a minimum positive energy separating the vacuum from the theory's excited states. The work also does not prove that the infinite-volume limit is unique.The reviewers added an important caveat: they used OpenAI's Astra to help examine the code and "do not claim to fully understand the proof yet".- Ends