Theoretical physics is usually a slow grind of chalk dust and coffee. You derive a formula, check it for months, and maybe publish. A new preprint suggests that rhythm is breaking. A team of top-tier physicists handed a solved problem about gluons to an AI model and asked it to solve the harder, gravitational version. The AI did not just copy the homework. It found a solution using a mathematical method the humans had not considered, effectively co-authoring a paper on quantum gravity.
Key Takeaways
- Single-minus graviton tree amplitudes are nonzero in the half-collinear kinematic regime.
- GPT-5.2 Pro derived the gravitational amplitudes using the directed matrix-tree theorem.
- Authors include researchers from IAS, Vanderbilt, Cambridge, Harvard, and OpenAI.
The paper addresses a niche corner of quantum gravity called scattering amplitudes. These are math tools used to predict what happens when particles smash into each other. For decades, textbooks said a specific configuration of gravitons—called “single-minus”—would always cancel out to zero at the simplest level. The new work proves this wrong.
Under very specific conditions where particle momentum aligns just right, these interactions actually exist. What makes this paper unusual is the method. The researchers gave GPT-5.2 Pro a previous paper on a different particle (gluons) and asked it to apply the logic to gravity. The model derived the correct formulas and drafted the paper.
The big deal
This matters for two very different reasons. First, it is a legitimate crack in the wall separating quantum mechanics from Einstein’s general relativity. The math connects to a symmetry theory proposed by Roger Penrose fifty years ago. Finding a concrete example of this symmetry in action helps physicists understand how gravity might work at the quantum level.
Second, it signals a shift in how science gets done. Usually, the hard part is the derivation. Here, the AI did the derivation using the “directed matrix-tree theorem,” a technique the humans found surprising and elegant. The humans spent their time checking the work rather than doing it. This suggests AI can now function as a theoretical collaborator rather than just a search engine or a calculator.
How it works
Physicists use scattering amplitudes to skip the messy middle part of a collision calculation.
Think of it like calculating a pool shot. A traditional calculation tries to track the ball rolling, hitting the rail, spinning, and hitting another ball. An amplitude calculation just looks at the starting position and the pocket the ball falls into. It ignores the journey to focus on the result.
In this case, the AI looked at the “start” and “end” for gravitons. It found that if the particles spin and move in a very specific formation, they do not cancel each other out. They interact. The AI used a complex matrix theorem to map these interactions, proving they follow a consistent mathematical pattern.
Gravitons: Theoretical particles that carry the force of gravity.
Helicity: The direction a particle spins relative to its motion.
The catch
This is theoretical work. It describes a “half-collinear” regime, which is a highly specific alignment of particles. It does not rewrite the laws of physics for everyday life. The paper is also a preprint. While the authors are from major institutions like Harvard and Cambridge, the work is part of an ongoing investigation.
The AI did not replace the physicists. The team notes that while the derivation was fast, they spent most of their time verifying the math and writing up the formal proofs. You still need an expert to know if the AI is right. The model also required a previous paper on gluons as an “anchor” to figure out the gravity problem. It did not invent the physics from scratch.
What to watch
Watch for the peer review process. Other physicists will now tear apart the derivation to ensure the “directed matrix-tree theorem” was applied correctly. Look for more papers using this “anchor” method. The team successfully transferred an idea from gluons to gravity by feeding the first paper to the model. Expect researchers to try this with other neighboring fields of physics.
Finally, keep an eye on the “w-(1+∞)” symmetry. If this result holds, it gives theorists a new tool to attack the problem of quantum gravity.
