Effects of attractions and social attributes on peoples’ usage intention and media dependence towards chatbot: The mediating role of parasocial interaction and emotional support

Abstract

PURPOSE: It is important to explore the relationship between humans and chatbots to improve human-robot interaction in the era of artificial intelligence. This study aims to explore the effects of attractions and social attributes of chatbots on users' media dependency and usage intention of chatbots, as well as the role of users' para-social interaction and emotional support gained from chatbots. METHODS: A total of 1,553 responses were collected based on a cross-sectional online survey. Utilizing the structural equation modeling approach, this study tested the relationships among exogenous variables (social attraction/task attraction, perceived competence/perceived warmth), endogenous variables (usage intention/media dependence), and mediating variables (para-social interaction/emotional support). RESULTS: The results show that the attraction and social attributes of chatbots, represented by ChatGPT, enable users to construct para-social interaction and obtain emotional support when chatting with them. Meanwhile, para-social interaction and emotional support can link users' perceptions of chatbots to their media dependency on and usage intention of them. This study provides theoretical and methodological references for examining the human-robot interaction relationship and offers insights into exploring the human-robot emotional connection. CONCLUSIONS: This study explores chatbots from the perspective of emotional connections, emphasizing how users' perceptions of chatbot attraction and social attributes facilitate para-social interaction and emotional support. Theoretically, it extends the application of para-social interaction theory and emotional support into the domain of human-chatbot communication, enriching the understanding of affective mechanisms in human-AI relationships. Methodologically, the study employs structural equation modeling (SEM) to test a multidimensional mediation model using large-scale survey data, examining psychological pathways linking chatbot characteristics to users' behavioral responses. These findings offer new insights for the optimization of human-computer interaction applications and the improvement of chatbot design in practice.