Stuart Russell answers 3+ questions in wake of Turing Lecture

In May 2020, CS Prof. Stuart Russell delivered the most highly attended Turing Lecture yet,  to a virtual audience of over 700 people from around the world, on the subject of provably beneficial AI.  In a follow-up article, "Three (plus) questions with Turing Lecturer Stuart Russell," he answers some of the many questions not covered during the live Q&A.  In his talk, Russell argues that "it is useful to imbue systems with explicit uncertainty concerning the true objectives of the humans they are designed to help. This uncertainty causes machine and human behaviour to be inextricably (and game-theoretically) linked, while opening up many new avenues for research."  The top three questions address how AI should make immediate choices, how to address changing preferences as society evolves, and how AI can be controlled to minimize bias.  The ideas discussed are explored in his most recent book, "Human Compatible: AI and the Problem of Control" (Viking/Penguin, 2019).  The Turing Lectures are hosted by The Alan Turing Institute, the UK’s national institute for data science and artificial intelligence, and should not be confused with the Turing Talks sponsored by BCS and IET.