Self-Disclosure to a Conversational Agent: The What and the How?

Abstract

Smart home assistants, or digital conversational agents (CA), can improve people’s lives by helping with reminders and schedules, as well as serving a social function. This study aims to assist with developing a CA for older people to support their cognitive and social needs by examining the user interaction. We evaluated the aspect of self-disclosure from the users from data collected in an earlier phase of the study with young college students at the University of Minnesota, where they participated in a series of interactions with a prototype CA. Results show that more varied content of individual's responses to the CA (i.e., variety in the specific type of information conveyed such as “work/study” or “leisure”) and greater variety in the form of individual's responses (i.e., variation in the way given content information is expressed, for example, “habit” and “judgments”, etc.) were associated with more user self-disclosure. These findings have implications for understanding and researching human computer interactions.