One of the deepest questions in the philosophy of science is: how do we generate hypotheses from data? The information we collect about the world contains only statistical correlations between observables, but the models we use to describe the world are causal. Our data can only tell us what happened, but our theoretical models can tell us what might have happened if we had acted differently. In this seminar, I will give a grand tour of thinking about causality in science, ranging across fields from philosophy to machine learning, all the way to time-travel and quantum gravity.