>In his recent book, Human Compatible, Russell lays out his thesis in the form of three “principles of beneficial machines,” echoing Isaac Asimov’s three laws o

>In his recent book, Human Compatible, Russell lays out his thesis in the form of three “principles of beneficial machines,” echoing Isaac Asimov’s three laws of robotics from 1942, but with less naivete. It is a recent trend to brush off Isaac Asimov's laws without an actual critique. Which betrays a lack of concrete thought in the matter. It's not that Asimov's laws are flawless, but that countering them seems easier than it really is. I have seen various bad, hand-wave dismissals but I can't recall any careful critiques. I have the new book, Human Compatible, and I buy Russell's argument but note that his rules are much more abstract, and therefore it is perhaps harder to counter.