AI3SD Video: How can explainable AI help scientific exploration?

Although models developed using machine learning are increasingly prevalent in scientific research, their opacity poses a threat to their utility. Explainable AI (XAI) aims to diminish this threat by rendering opaque models transparent. But, XAI is more than just the solution to a problem--it can also play an invaluable role in scientific exploration. In this talk, I will consider different techniques from Explainable AI to demonstrate their potential contribution to different kinds of exploratory activities. In particular, I argue that XAI tools can be used (1) to better understand what a "big data" model is a model of, (2) to engage in causal inference over high-dimensional nonlinear systems, and (3) to generate algorithmic-level hypotheses in cognitive science and neuroscience