¹⁷This division into two kinds of uncertainty could be criticized on philosophical grounds. Laplace proposed that an intellect (called a “demon” by later commentators) which knows everything about the world would be able to perfectly accurately predict everything, and nothing would be uncertain to it. Thus, from this viewpoint, uncertainty is always a reflection of ignorance about some aspects of the world. I shall not go into that debate here, and just acknowledge that the division I make here is not very rigorous while in line with how randomness is often conceptualized in AI theory. This is basically the distinction between “aleatoric” and “epistemic” uncertainty discussed, e.g., by Hüllermeier and Waegeman (2021); Charpentier et al. (2022); Lockwood and Si (2022).