Building Trust over Time in Human-Agent Relationships

Abstract

This paper aims to understand how long-Term trust and distrust develop between humans and agents (smart objects). We first conducted a qualitative study to explore key factors that lead to trust and distrust, how the human-Agent trust journey develops, and what roles these trust-building factors play in the journey. This qualitative study involved an open-ended questionnaire completed by 621 participants and a five-day diary study with 60 participants that examined 499 human-object and human-Agent relationships. Next, we conducted a mixed methods study with 146 participants to rate the importance of key factors identified through analysis of the data collected in the first study. We contribute to the HAI community by identifying eight factors accounting for participants' trust and distrust toward new objects, how these factors play different roles at different phases of the human-Agent trust journey, and how important these factors are for both trust and distrust. We identified ebbs and flows in the human-Agent trust journey over time, revealing time periods when trust is particularly vulnerable. We also found the most important factors for building trust and avoiding distrust did not entirely overlap. We discuss these findings and their implications for designing agents which need to foster trustful long-Term relationships with humans.