Room: Turchese

9:30-9:35 Welcome and introduction
Neil Lawrence, University of Cambridge, Jessica Montgomery, University of Cambridge

9:35-10:15 Keynote: Animal Behaviour as Algorithms
Cait Newport, University of Oxford

10:15-10:30 From Hype to Hypotheses: Why AI-for-Science Is Hard - and What It Takes to Make It Work
Tegan Emerson, Pacific Northwest National Laboratory

10:30-10:45 Scientific Inference with Diffusion Models
Stephan Mandt, University of California, Irvine

10:45-11:00 Benchmarking for Experiments in Inference across Dynamic Spatial Boundaries: Water Quality
Claire Hardgrove

11:00-11:30 Coffee

11:30-13:00 Working Session 1: What scientific problems could become solvable with AI in the next 5 years and what AI advances do we need to get there?

13:00-14:30 Lunch

14:30-14:45 Fireside chat on the Raspberry Pi for Scientific AI Agents
Katherine Gorman (Moderator), Neil Lawrence, University of Cambridge

14:45-15:15 Working Session 2: Design the Raspberry Pi for Scientific AI Agents

15:15-16:15 Panel discussion: Are we heading for a paradigm shift in science through AI?
Ryan Daniels, University of Cambridge, Arno Solin, Aalto University, Lydia France, University of Oxford & Alan Turing Institute, Stephanie Hyland, Microsoft Research

Abstracts

Keynote: Animal Behaviour as Algorithms

My research seeks to understand how animals perceive their environment, how they transform sensory input to behavioural output, and how they flexibly adapt their behavioural strategies in response to changes in the environmental or context. I am currently focused on understanding how coral reef fish can use visual information to navigate, and how they can move through complex terrain even under low visibility conditions. By identifying the navigational mechanisms that underpin this behaviour, I aim to find lightweight and efficient navigation algorithms that can be applied to underwater autonomous vehicles. In this talk, I will share my research on the visual and cognitive abilities of fish, demonstrate how advances in AI are expanding the scope of questions behavioural biologists can tackle, and explore how insights from natural intelligence might inspire the design of artificial systems.