Future Directions in Artificial Intelligence in Fundamental Physics

by Lukas Alexander Heinrich (TUM)

Tuesday 3 Feb 2026, 16:00 → 17:00 Europe/Berlin

ENC-D308

Particle Physics and related fields are scientific domains that are amongst the earliest adopters of machine learning in the sciences. This is not coincidence: our fields rely fundamentally on stochastic generative modelling with simulators and pattern recognition and thus are naturally very aligned to the strengths of ML/AI. It therefore may be surprising that there has not been yet an AlphaFold like field-wide breakthrough in fundamental physics and e.g. the leading ML applications in particle physics (with millions of parameters) still focus on narrow tasks while foundation models like ChatGPT have billions of parameters. In this talk I will discuss recent work towards building large-scale AI models for fundamental physics and touch on prospects of near-autonomous agentic systems for future frontier experiments.