Embodied Conversational Agent Teams for Trust Calibration in Automated Vehicles

Representing sub-systems of automated vehicles with a team of conversational agents has been suggested in the past to improve the process of trust calibration. Experiments so far did not include embodied agents. In this work, we present two potential analogies derived from a company hierarchy and a spaceship crew. We developed comic strips showing everyday interactions with these agents in an automated vehicle and evaluated them in a Vignette study in comparison with single-agent approaches. Our results suggest that a multi-agent approach could be beneficial for trust calibration, but only for the company and not the spaceship analogy. Only some of the design decisions regarding the agents' personalities worked out as intended. Still, we believe the concept is promising for future investigations.