???? UNDER THE LAMPLIGHT-

What would happen if we succeed in building a superhuman machine intelligence? In his book, Human Compatible: AI and the Problem of Control, artificial intelligence expert Stuart Russell argues that current approaches to AI are dangerously misguided and proposes a new, more humane direction for the field. Russell advocates for what he calls “safe intelligent systems,” which are designed not just to be intelligent but also to be safe and beneficial for humanity. He suggests a number of rules for making these kinds of systems, such as giving AI systems access to a wide range of information about the world and humans’ place in it, making them so that they always ask humans what to do when they don’t know what to do, and giving them human-like social skills so that they can communicate and work well with humans.