A team of researchers has used artificial intelligence (AI) to discover a simple formula for a type of particle interaction that was previously thought to be impossible at its simplest level. The study, published on the pre-print server arXiv, shows that when particles are arranged in a specific way, their interaction does not disappear and can be built step by step using a standard method in particle physics.
Using an AI system called GPT-5.2 Pro, the researchers were able to identify a pattern and conjecture a general formula for the interaction, which was then rigorously proven and independently checked by human scientists. The result has significant implications for theoretical physics, particularly in understanding the internal structure of scattering amplitudes.
The study focuses on gluons, particles that carry the strong nuclear force, and uses a concept called scattering amplitude to encode the probability of particle interactions. The researchers show that when certain conditions are met, the scattering amplitude does not vanish, but rather exists on a precisely defined slice of momentum space.
The AI-assisted discovery is significant because it highlights the potential for machine learning algorithms to assist human scientists in generating new insights and simplifying complex calculations. The researchers believe that this approach may be particularly well-suited for domains of theoretical physics where hidden simplicity inside complex algebra can be identified.
While the study applies to tree-level amplitudes and specific kinematic regimes, it has implications for more general cases, including gravitons and supersymmetric extensions. Further research is needed to fully understand the structural role of these amplitudes within broader theories.
This discovery marks an exciting development in the integration of AI into scientific research, particularly in theoretical physics. As Professor Nima Arkani-Hamed said, “Finding a simple formula has always been fiddly, and also something that I have long felt might be automatable by computers. It looks like across a number of domains we are beginning to see this happen; the example in this paper seems especially well-suited to exploit the power of modern AI tools.”
Source: https://thequantuminsider.com/2026/02/13/ai-scientist-spots-what-physicists-missed-in-gluon-scattering