The effect of knowledge-based system assistance on piloting performance, workload, and satisfaction

A prototype software system has been built to evaluate the usefulness of simple artificial intelligence systems in aircraft cockpits. Laboratory evaluations of the system were conducted using a simulated aircraft in a full-mission scenario. Pilot task performance and workload were measured under two conditions: baseline and enhanced with the prototype knowledge-based decision-aiding system; some of the results achieved statistical significance. On the average, the knowledge-based system improved pilot performance in locating surface objects by 36%. Performance was improved an average of 42% for tasks related to successful response to mission emergencies. Workload was decreased by 13%. Overall, 83% of the pilots perceived the prototype system to be as effective as, or more effective than, the baseline system. In direct comparisons 81% of the pilots preferred the enhanced to the baseline system.<
>