Adaptive Computerized Training System (ACTS): Relationships Between Utility Similarity and Strategy Similarity

摘要

The results of an effort to apply Artificial Intelligence (AI) techniques to electronic troubleshooting training was the Adaptive Computerized Training System (ACTS). ACTS is an 'intelligent' simulator on which a student troubleshoots a complex electronic circuit by making various test measurements, replacing the malfunctioning part, and making final verification measurments. The 'intelligent' component is an adaptive computer program which models the student's decision structure, compares this structure to that of an expert, and when complete, will adapt the instructional sequence and provide feedback to eliminate discrepancies between the two. An Expected Utility (EU) model of decision making is the basis of the models of both the student and expert troubleshooters. A strong positive relationship between utility similarity and strategy similarity was obtained. This is a prerequisite to using discrepancies between student and expert utilities as a basis for adapting training. Because the experimental conditions were somewhat artificial, the strength could be expected to be attenuated in a 'real world' training situation.