The Influence of Attachment Style Variations on Problematic Use of AI Chatbots: A Mediation Study of Social Anxiety and Human-Robot Empathy

Abstract The rapid adoption of AI chatbots has brought about the rise of emotionally intimate human–machine relationships.This study investigates how attachment style, as an individual difference factor, influences problematic use of AI chatbots through the mediation effects of social anxiety and human–robot empathy. Drawing on data from 534 valid questionnaires, this study found that anxious attachment significantly heightened both social anxiety and human–robot empathy, thereby indirectly contributing to problematic use of AI chatbots Avoidant attachment, by contrast, affects problematic use primarily via social anxiety, with no significant association with human–robot empathy. In addition, social anxiety and human-robot empathy served as chain mediators in both attachment paths. The results show that in the process of using chat behavior in a problem-solving manner, social anxiety acts as a precursor mechanism for emotional triggering, prompting individuals to avoid real-world social interactions and direct their emotional needs toward AI chatbots; human-robot empathy serves as a safeguard mechanism for emotional response and psychological compensation, maintaining individuals' emotional engagement during interactions. These two factors work synergistically to foster increased engagement and ultimately contribute to the development of problematic use patterns. This study revealed the underlying emotional pathways and psychological mechanisms behind human-AI interactions, offering a new perspective for understanding emotional behaviors in the digital era.