ABSTRACT
An analysis of AI programs reveals that there exists a scope of massive parallelism in various phases of reasoning and search. For instance, the search problem in a given state-space can be subdivided and allocated to a number of processing elements. Further, many modules of the reasoning programs can also be realized on a parallel and distributed architecture. Generally, the parallelism in a reasoning program can appear at the following three levels: the knowledge representation level, the compilation and control level and the execution level. The chapter provides a brief introduction to the architecture of intelligent machines with special reference to representation and execution level parallelism in heuristic search, production systems and logic programming.
