Computer simulation vs. machine learning – a philosophical comparison

Digital computers have had a profound impact on modern physics. In this talk, I discuss two computer-based methods, viz. computer simulation and machine learning. To have a common ground for the comparison, I assume that both methods are used to infer hitherto unknown information, typically about future states of a system. Important aspects of the comparison are the power to achieve various epistemic tasks, the ways in which results are justified and the opacity of the methods.