Issues in distributed artificial intelligence

This paper addresses some of the issues involved in fomalizing AI concepts to deal with several artificially intelligent agents cooperating to achieve a system-wide goal. The suggestion is made that the approaches used to construct AI systems which are operating autonomously in a single domain environment are not necessarily the same approaches that should be considered in building a set of intelligent agents that are physically or logically separated and are trying to achieve a common global goal through the use of individual subgoals and subplans. Issues addressed concern the difficulties of formulating temporal relationships, difficulties of interfacing conflicting or concordant goals, and communication between subplans. A simple distributed problem is also presented.