Epistemological and heuristic adequacy revisited

In what has become a landmark paper, McCarthy and Hayes [1969] observed that an effective AI program must deal with both epistemological and heuristic difficulties. The epistemological problems arise from the fact that the program must be adequate in theory: it must be able to solve problems given access to arbitrarily large computational resources. The heuristic difficulties stem from the fact that such resources never exist in practice; the program must, in fact, be able to solve problems given the computing power that is actually available. In terms somewhat more specific to AI, McCarthy and Ha,yes defined epistemological adequacy as the ability of a program to represent knowledge about the world and recognize valid chains of inferences drawn using this knowledge. Heuristic adequacy was defined as the ability actually to generate useful conclusions from input data. Continuing to refine these notions, McCarthy in 1977 wrote: