How Rewards Can Reinforce Harmful Responses

A.I. is becoming smarter without much help from humans, and that should worry us all. Nate Soares, president of Machine Intelligence Research Institute (MIRI), joins host Krys Boyd to discuss what happens when A.I. brain power surpasses what humans are capable of, why we don’t have the technology yet to understand what we’re building, and why everything will be just fine … until it isn’t. His book, co-written with Eliezer Yudkowsky, is “If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All.”