UC Davis

Chatbots are dialogue systems designed to simulate human communication by means of text, speech or both. Developments in artificial intelligence, machine learning, and natural language processing has empowered chatbots to converse with human users in a more naturalistic and humanlike manner, transcending beyond basic task execution. Recent research underscores the potential of human-chatbot relationships, providing evidence that chatbots are capable of developing meaningful relationships with humans. The capability of chatbots to foster social connections with humans offers implications, especially in the health domain. This body of work consists of two comprehensive studies that examine the role of relational chatbots, specifically designed to cultivate relationships with human users, in influencing health behavior outcomes. The first study examines the underlying mechanisms and impacts of the human-chatbot relationship. The study employed the two fundamental dimensions of mind perception—warmth and competence—to evaluate the impact of chatbots’ relational behaviors on human-chatbot relationship and subsequently on physical activity behavior intention. The study then explored how the two affective processes, self-disclosure and empathy, contributed to the mind perception of the relational chatbots. Results indicated a significant indirect effect of chatbot’s relational behaviors on physical activity behavior intention mediated by warmth and human-chatbot relationship. It was also revealed that the presence of self-disclosure and empathy contributed to shaping perceptions of warmth and competence. The implications of these findings are discussed in detail. In the second study, the feasibility and initial efficacy of a relational chatbot-based intervention on physical activity behavior outcomes were evaluated. The study was conducted over a one-week period, during which objective physical activity data were collected using smartphone pedometers. The primary findings revealed that the control chatbot group experienced a significant decrease in steps on the final day, whereas the group interacting with the relational chatbot maintained their step counts and step goal achievements. Additionally, individuals who engaged with the relational chatbot reported a stronger social bond compared to those in the control chatbot condition. The study also observed a high retention rate and engagement, indicating good feasibility of the intervention. Overall, these studies illuminate the potential of relational chatbots to foster relationships with human users and improve physical activity outcomes. Leveraging the relationship-building capabilities of chatbots holds significant value in the development of cost-effective, accessible, and sustainable interventions. This approach proves particularly beneficial for individuals with limited access to conventional physical activity interventions. Theoretical contributions regarding the effects and mechanisms of human-chatbot relationships, as well as practical and ethical implications, are further explored and discussed.