The search horizon is the part of the {[search space}} that has been explored at a particular decision point. In a simple problem it is often possible to explore the entire search space of possible solutions. However, in others this is not possible, for example there and more than1040 legal chess board positions and more than 2×10170 {Go}} board positions. In planning algorithms, such as route planning there are often uncdertainties whuch expand this space further. In such situatiosn, only a portion of the space can be explored before a decsion has to be made. In some cases, this isfixed, for exampel, looking exactly two moves ahead in a chess game, but, given the limited total time available, the search horizon is ofetn of veriable depth.. Heauristics are particularly important in choosing which portions of the search space to explore more deeply. In systems where AI decisonis lead to an action being taken, say a move in chess, there will be some form of response, the opponenst move, which reduces the possible future search space and means that at the next step the search horizon can extend more deeply in some areas. This is also the case where there is no opponent, for example a robot moving will mean that it can perhaps see around a corner reducing uncertainty and pruning the search space. In soem occasions the system may perform an epistemic action, that is a deliberate choice to take an action that reveals information.
Used on pages 78, 226, 228, 229