rather than using purely boolean logic use a heuristic function to evaluate the appropriateness of each option given the moment to moment situation in the game and base their decision on the resulting score.
the approach typically either take the option with still highest score or use weight based random to pick a way that is nondeterministic but still gives a greater chance of being selected to the best options.
introduced the idea dual utility reasoning this combines absolute utility is used to divide the options into categories and ensure that only options from the most important or most relevant category will be selected.
relative utility is used to pick random from among the options withing the category, this helps to prevent both the predictability of absolute utility and the occasional poor choices of relative utility.
it provides a bit more flexibility and expressiveness to the games design when are responsible for configuring the ai
