How Conversational Contingency Shapes Trust in Human-AI Interaction: Evidence from a Two-Wave Mediation Study

Conversational contingency—the alignment of a chatbot’s responses with users’ prior inputs—is central to meaningful interaction, yet its implications for trust in AI remain underexplored. Drawing on two-wave panel data from Chinese chatbot users (N=391), this study examines how perceived conversational contingency is linked to user trust, both directly and indirectly through social presence and user engagement. Findings indicate that when chatbots demonstrate contextual responsiveness, users report heightened perceptions of social connection and deeper cognitive and emotional involvement, which are both associated with increased trust. This study extends existing frameworks of human-AI interaction by articulating the relational pathways through which interactional quality shapes user perceptions. The results also suggest that enhancing conversational contingency can serve as a key design strategy for promoting more trustworthy and socially engaging AI communication systems.