Heuristic is a practical rule-of-thumb that helps an algorithm or designer choose promising options without exhaustively testing every possibility. In game AI, heuristics are central to search and planning: A* uses a heuristic estimate of remaining distance to guide pathfinding; tactical AI uses heuristics to rank cover, threat, or opportunity; ML systems may use heuristic features before any learned model is added. A heuristic is useful because it is fast and informative, not because it is guaranteed to be correct in every case. (Bhargava, Grokking Algorithms, see source-grokking-algorithms; Millington, Artificial Intelligence for Games, see source-artificial-intelligence-for-games)
Good heuristics reduce search cost while preserving acceptable behaviour. Bad heuristics bias the system toward locally plausible but globally poor choices.
Related: pathfinding-algorithms, goal-oriented-action-planning, mcts, machine-learning-games