Beyond the Halo: Investigation of Trust and Functional Specificity in Automated Driving with Conversational Agents

Abstract

Only a few works so far have addressed functional specificity for trust formation in automated driving. Previous research indicated that drivers could hardly distinguish between sub-systems, while their trust is influenced by other in-vehicle technologies. Thus, we conducted a user study where participants had to supervise a level 2 vehicle while reading and communicating with conversational agents. In two conditions, the vehicle was either represented by a single agent or by two agents that portrayed the driving automation and the infotainment system. We hypothesized that a clear differentiation between sub-systems could allow drivers to better calibrate their trust. However, our results show quite the opposite. Correlation analyses suggest that participants’ functional specificity was high, and they based their situational and general trust ratings mainly on the perception of the driving system. Also, dispositional trust did not influence trust formation, but many participants still failed to monitor the system appropriately.