Two graduate students at a university in California have built an artificial intelligence system that can dream up new explanations for one of science's most puzzling mysteries: why tiny particles called neutrinos have any mass at all.
Victoria Knapp-Pérez and Jake Rudolph, doctoral candidates at the University of California, Irvine, created a computer program named AMBer — short for Autonomous Model Builder. Unlike typical AI that learns from existing data, AMBer uses a technique called reinforcement learning, which means it teaches itself by trying things, seeing what works, and trying again.
"AMBer's RL framework allows it to learn about the space of theoretical models as it explores, effectively creating its own training data as it searches for promising models," Rudolph explained.
Neutrinos are incredibly light particles that pass through matter — including your body — by the billions every second. But here's the puzzle: the Standard Model, physicists' best theory of how matter works, says neutrinos should have no mass at all. Yet experiments show they do have a tiny amount of mass. Nobody knows why.
Knapp-Pérez and Rudolph's AI system builds theoretical models by picking mathematical rules called symmetry groups, deciding which particles to include, and setting how those particles behave. It then tests each model against real experimental data while trying to use as few adjustable numbers as possible — fewer numbers means a more powerful, predictive theory.
The researchers first tested AMBer on well-known neutrino theories and showed it could reproduce known results. Then they pointed it at unexplored mathematical frameworks. The system came back with entirely new candidate models that human physicists had never considered.
The team published their findings in the journal Communications Physics. Their work isn't meant to replace scientists, though. It's designed to do the heavy lifting of sorting through billions of possible theories and highlight the most promising ones for human researchers to investigate.
"AMBer functions as a filter, giving human physicists a better-informed starting point from which to study more complex behavior of neutrino models," Knapp-Pérez said.
The neutrino mass problem has puzzled physicists for decades. If AMBer can help narrow down the possibilities, it might eventually lead to an answer — or at least point scientists in the right direction.