In disaster films, the conflict between humans and intelligent machines is often depicted as inevitable, with a depressing outcome. Stuart Russell, one of the leading researchers in the field of artificial intelligence (AI), argues in this groundbreaking book that we can avoid such a scenario – but only if we fundamentally rethink the concept of AI. He presents an overview of the groundbreaking discoveries that still need to be made on the path to superhuman AI. However, if these discoveries occur and superhuman AI emerges, entities much more powerful than humans will appear. How can we ensure that they do not dominate us? Intelligent machines can ominously take control of decision-making in the real world and pursue their own goals derived from their programming. In this publication, Russell introduces the concept of "provably beneficial" AI, whose algorithms can be mathematically proven to pursue goals beneficial to their human users. The Guardian therefore labeled the book 'Human Compatible' as one of the most important texts on artificial intelligence.
