Stuart Russell

Stuart Russell wrote "Artificial Intelligence: A Modern Approach" with Peter Norvig in 1995. It is the standard AI textbook, now in its fourth edition, used in over 1,500 universities. In terms of how the field describes itself and what it teaches to new practitioners, no single work has had greater reach.

Russell's research shifted significantly in the 2010s toward AI alignment — the problem of ensuring that AI systems pursue the goals humans actually want rather than proxy metrics. His 2019 book "Human Compatible: Artificial Intelligence and the Problem of Control" articulates what he calls the "standard model" problem: conventional AI design assumes the machine is given a fixed objective to maximise, but it is impossible to specify all of human values precisely in a reward function. He proposes an alternative foundation — AI systems that are uncertain about human preferences and designed to ask for human input rather than act autonomously on assumed objectives.

Russell testified before the US Senate and has engaged with UN processes on lethal autonomous weapons. He co-founded the Center for Human-Compatible AI at Berkeley. His technical standing — he is among the most cited AI researchers in the world — gives his safety arguments different weight than those of ethicists or policy researchers who lack equivalent technical credentials.