GRASP on Robotics is an inaugural series of talks hosted by the GRASP Laboratory. GRASP leverages academic, research, and industry connections to deliver a set of high class tech talks with the mission of providing technical topics and meaningful discussions.
Join the live-stream here Fridays from 10:30am -11:45am followed by a Q&A panel between our speaker, faculty, and students until 11:45am.
[VIRTUAL SPEAKER]: Fall 2025 GRASP on Robotics: Max Welling, University of Amsterdam & CuspAI, “On the Role of Uncertainty in Automating the Scientific Process with AI”
This event will have a VIRTUAL SPEAKER. Streaming will be in-person only in Wu and Chen.
ABSTRACT
We are in the middle of a transformation of the scientific process. AI is now rapidly automating scientific discovery by accelerating scientific simulation and optimisation. For example, we can now shortcut expensive quantum mechanical calculations such as DFT with much faster machine learned interatomic potentials and forces, numerically “solve” PDEs with neural networks, or accelerate experimental design with Bayesian optimisation. However, in both applications of AI the role of uncertainty is key because ML models can only be trusted in regions of input space where it has seen data. Extrapolating too far away from the data runs the risk of a large prediction error that goes unnoticed. In this talk I will first provide a high level overview of the increasingly important role of AI in science, and then proceed with introducing a new efficient Variational Bayesian method for uncertainty quantification. I will show the effectiveness of this method on PDE solving and molecular generation (BARNN) and on ML interatomic potentials and forces (BLIP).