New Tools for Old Problems: Generative AI for Discovery in Fundamental Physics

David Shih surveys generative AI applications to discovery in fundamental physics. Normalising-flow methods enable model-agnostic anomaly searches at the LHC and identification of stellar streams and faint galaxies in astronomical data. Generative inference also supports pulsar-timing gravitational-wave analysis and, combined with physics-informed networks, measurements of local dark-matter density. The talk concludes with agentic language models and new forms of collaboration between researchers and AI.