Adventures with Agentic AI in Particle Physics

by Prof. David Shih (NHETC, Rutgers University)

Tuesday 18 Aug 2026, 16:00 → 17:00 Asia/Shanghai

Tsung-Dao Lee Institute/N6F-N600 - Lecture Room (Tsung-Dao Lee Institute)

Abstract:

Agentic AI -- where powerful LLMs are given the ability to autonomously plan and execute complex tasks -- is reshaping nearly every corner of society. I will give an overview of the impact agentic AI is having in high energy physics, with examples drawn from my recent research. On the one hand, many groups are exploring the development of fully automated agentic AI systems for anomaly detection, data analysis, and other important tasks in particle physics. I will describe ColliderBench, a recent paper from my group at Rutgers, which provides a quantitative benchmark for these systems based on the highly nontrivial task of reproducing an analysis paper from the LHC. On the other hand, some of us are exploring the frontier of human-AI collaboration, where the agent is treated like a remote graduate student that does all of the hands on work in a research project. I will describe my recent experience working in this "vibe physics" mode with Claude Code, which led to two single-author papers developing a new framework for symbolic simplification of mathematical expressions based on an analogy with scrambling and unscrambling Rubik's cubes. Finally, I will conclude with some lessons learned working with agentic AI and general thoughts for the future of AI and physics.

Speaker Bio:

David Shih is a theoretical particle physicist currently working at the interface of AI/ML and fundamental physics. Educated at Harvard University (AB), University of Cambridge (M.Phil), and Princeton University (Ph.D), Shih has been a Professor at the New High Energy Theory Center in the Department of Physics & Astronomy at Rutgers, New Brunswick since 2010. Shih's career has spanned a broad set of topics, from neutrino experiment and X-ray astronomy, to string theory, supersymmetry, collider phenomenology and dark matter. His work has been recognized by numerous awards, including the DOE Early Career Award, the Sloan Foundation Fellowship and the Friedrich Wilhelm Bessel Research Award from the Alexander von Humboldt Foundation.

Alternative online link: https://meeting.tencent.com/dm/kphUOa4jGdhD
ID: 762219102 (No Password)