The ultimate goal of the science of artificial intelligence (AI) is to establish programs that will use algorithmic computer techniques to imitate the heuristic thought processes of humans. Most AI programs, especially expert systems, organize their knowledge into three specific areas: data storage, a rule set, and a control structure. Limitations restricting the development of AI include the use of limiting heuristics, data processing speed, organization of memory, AI accuracy, and sensitivity (underdeveloped recognition and processing of data subtleties). AI's most immediate future impact will be in the analysis phase of the Training Technology System, a five-phase model of the training process. The two main components of this phase in which expert systems could be applied in the future are the processes of training needs assessment and work behavior analysis. (YLB)