Bounded Optimality
The notion of optimizing not the action that is taken but the algorithm that is used to choose that action. In other words, taking into account the cost of making a decision, and using resources to think rather than act.
EverydayConcepts.io
Origin
Eric Horvitz introduced the term in his 1989 Stanford doctoral dissertation Computation and Action Under Bounded Resources, proposing that intelligent agents be evaluated on program-level optimality within real hardware constraints. Stuart Russell and Eric Wefald extended and formalized the idea in their 1991 MIT Press book Do the Right Thing: Studies in Limited Rationality, establishing it as a foundational concept in AI.
Updated February 22, 2026