An Exploratory Study on the Impacts of Voice-Based Conversational Agents with Proactive Interactions in the Driving Context

Advanced artificial intelligence (AI)-based technologies offer great opportunities for enhancing the driving experience. Proactive voice agents (PVAs) present benefits by being proactive rather than responding to requests. This study uses a driving simulator to investigate whether PVAs can create value for drivers. We designed three types of PVAs: a task agent, a social agent, and a companion agent, as well as a control agent for a 2 (small talk vs. non-small talk) (Formula presented.) 2 (task-based talk vs. non-task-based talk) between-subjects experiment. We found that PVAs (i.e., social or companion agents) engaging in small talk increased perceived intimacy, but task agents increased perceived pressure. Moreover, experienced drivers preferred PVAs that engage in small talk to avoid sleepiness or boredom, while novice drivers favored control agents. As younger, experienced drivers have the highest technology acceptance, they will be potential adopters of PVAs.