Human Compatible: Artificial Intelligence and the Problem of Control

Many people fear that advanced AI would wipe out human occupations and social interactions and eventually destroy human civilisation. The outcome of a war between humans and machines is widely assumed to be predetermined. In this ground-breaking book, renowned AI expert Stuart Russell believes that this outcome is avoidable, but only if we completely reimagine AI. Russell begins by investigating the nature of intelligence and how it manifests in both people and robots.

From intelligent personal assistants to greatly accelerated scientific research, he highlights the benefits we may expect in the near future and the AI advances that still need to occur before we reach superhuman AI. He also details how people are already abusing AI in harmful ways, for as with autonomous weapons or cyberattacks.