OpenAI outlines lattice construction for the full Standard Model
OpenAI has proposed a candidate nonperturbative definition of the Standard Model on a spacetime lattice, with GPT-6 Astra helping develop the construction and detailed proof. The work targets a longstanding problem in particle physics: describing the full theory without treating interactions as small corrections.
- OpenAI proposed a candidate nonperturbative definition of the Standard Model on a lattice.
- GPT-6 Astra helped turn the strategy into a construction and developed the detailed proof.
- The Standard Model’s chiral weak force creates a lattice problem known as doubling.
- Lattice field theory can support direct computer calculations without perturbation theory.
A missing foundation for the Standard Model
The Standard Model describes known elementary particles and three fundamental forces: the strong force, the weak force and electromagnetism. OpenAI says the theory has matched experiments across eighteen orders of magnitude, from hydrogen’s 21-centimeter radio emission to collisions at the Large Hadron Collider.
Despite that success, physicists do not have a nonperturbative definition of the full theory in continuous spacetime. In perturbation theory, interactions are treated as a series of small corrections to a simpler problem in which they are turned off. That method can produce highly accurate results, including a prediction of the electron’s magnetism to within a few parts in a trillion, according to OpenAI.
But the method does not capture all of the Standard Model’s physics. OpenAI uses the proton as an example. Its mass comes primarily from the motion and interactions of its quarks and the gluons associated with the strong force. Turning off that force causes the proton to fall apart, so perturbation theory cannot provide a useful guide to the proton’s mass.
Why use a lattice
A lattice replaces continuous spacetime with a grid of points. The grid spacing sets the shortest distance the theory can resolve, while a finite grid provides a rigorous nonperturbative definition at a given spacing.
The approach also makes it possible to put the theory on a computer and calculate predictions directly, without relying on perturbation theory. OpenAI says lattice field theory has already advanced understanding of the strong force, including a quantitative, first-principles understanding of the proton’s mass.
The obstacle to extending this method to the entire Standard Model is that the theory is chiral. Its weak force treats left-handed and right-handed fermions differently, unlike the strong force and electromagnetism. A straightforward lattice formulation of a left-handed fermion naturally produces a right-handed partner, creating the wrong particle content for the weak force. This problem is known as doubling.
GPT-6 Astra helped develop the construction
In an October preprint, OpenAI proposed a candidate lattice definition. The company says it had a strategy for addressing the problem but had not proved that the approach worked in all the required cases. GPT-6 Astra helped turn that strategy into a construction and took the lead in developing the detailed proof.
OpenAI explains doubling through an analogy with light. A box with ordinary mirrors can turn left-circular light into right-circular light when the light reflects, meaning that a setup intended to contain one polarization ends up containing both. Special chiral mirrors can avoid that outcome by reflecting one handedness while allowing the other to pass through.
The source compares this optical solution with lattice rules developed decades ago by Paul Ginsparg and Nobel laureate Ken Wilson. It also names Rajamani Narayanan and Herbert Neuberger in the discussion of the lattice approach, although the supplied source ends before completing that account.
What the work could enable
According to OpenAI, the proposed construction is a step toward defining the full Standard Model nonperturbatively on a lattice. Such a definition would give researchers a framework for calculating predictions with the interactions present from the start, rather than treating them as small corrections.
The source presents the work as a response to a foundational gap that has remained for more than fifty years, while also making clear that the announcement concerns a candidate definition and its proof strategy. Its significance lies in addressing the chiral nature of the weak force, the feature that has made a complete lattice formulation especially difficult.
OpenAI: A step toward defining the Standard Model on a lattice
