Reducing Overtrust in Failing Robotic Systems

摘要

In general, people tend to place too much trust in robotic systems, also in emergency situations. Our study attempts to discover ways of reducing this overtrust, by adding vocal warnings of error from a robot that guides participants blindfolded through a maze. The results indicate that the tested vocal warnings have no effect in reducing overtrust, but we encourage further testing of similar warnings to fully explore its potential effects.