My final lecture will be about some of the criticisms of AI. I need to prepare for that by talking about the limits to computation: what computers can't do. Of course, the average computer can't transmute lead, generate antigravity, or even make toast. What we're interested in is restrictions on the type of computation that can be carried out.
I'll start with two examples of limitations on neural nets, and how they imply the idea of limits to what a learner can learn. Then I'll move to the most general kind of computer known, and what it can or can't do.