I will confess that here are certain sections of "Human Compatible" that are, despite the book's accessible tone, still beyond me; with that said, I think this book posits some interesting ideas (including the one which might be summarized as "everything we think about A.I. is fundamentally wrong" in that we have historically misapplied the definition of intelligence itself). The early chapters serve as a great introduction to the topic for those of us who have not brushed up on our quantum computing advances in a while, and the later chapters move into rewriting the underpinnings of A.I. theory. I rather like Russell's take on the subject, and I think it makes for a great counterpoint (or antidote) to the fears of Terminator-style machine takeovers.