Physics & Astronomy Colloquium (In-Person): Dr. Falk Herwig, University of Victoria

About

Elliott (ELL) 168. “ Machine Intelligence (MI) in the Loop: A Physicist’s Research and Teaching Relationship with MI Agents ” Agentic AI, or better yet MI, has moved from answering questions to carrying out long-horizon scientific work: strategizing how to reach a goal, writing code and running simulations, driving long multi-step investigations with complex tool use, analysing results, and drafting the paper. I will show what these agents actually are, how they differ from the traditional computing, and how our group runs research projects with agents working days (and nights), under human supervision. From there I will step back to the larger question this raises for the university, in both research and teaching, and argue for a posture of directed engagement in which we set the intent and purpose, and thereby generating a counterpoint to AI uses not aligned with our values. I will close with a concrete road map for bringing this into our own courses through Vela, a governed platform for MI augmented teaching and learning, we are building and testing at UVic.