Abstract: In 1951, Alan Turing predicted the eventual loss of human control over machines that exceed human capabilities. I will argue that Turing was right to express concern but wrong to think that doom is inevitable. Instead, we need to develop a new kind of AI that is provably beneficial to humans. I will describe an approach — assistance games — that seems promising. On the horizon, however, are a number of open questions, some of them familiar to moral philosophers and government regulators and some of them new.

A joint event with the Philosophy, Law & Politics Colloquium Series, funded by the Hewlett Foundation and the Institute for Ethics in AI Colloquium Series.