A fundamental problem in artificial intelligence research, first identified by John McCarthy and Pat Hayes (‘Some Philosophical Problems from the Standpoint of Artificial Intelligence’, Machine Intelligence, 1969). A common-sense appreciation of the world includes a great many automatic constraints on belief. For example, if a proposition p is true at one time, and event e occurs, we just know whether e is relevant or not to whether p is true at a later time. We ignore beliefs that are obviously irrelevant to our goals and we keep track of salient changes, despite the holism of change, or the fact that virtually anything can affect anything else. The frame problem is that of programming ‘frame axioms’ to reproduce this ability in any remotely realistic device supposed to mimic human intelligence. See also narrative competence, Turing test.
Philosophy dictionary. Academic. 2011.