The assumption that an AI agent will act in a way that maximizes its expected utility based on its understanding of the environment.
Generality: 775