Why do people prefer social sharing of emotion with conversational AI over human partners? A structural topic modeling approach

Abstract

The social sharing of emotion, traditionally studied in human-to-human interactions, is increasingly occurring with conversational AI systems. However, the specific motives for preferring AI over human partners for this purpose remain relatively underexplored. The present study aims to identify these motives and examine their associations with individual differences—namely, frequency of conversational AI usage, social anxiety, and self-esteem. Open-ended survey responses from 178 university students, describing situations in which they would prefer to share emotions with conversational AI rather than with human partners and their reasons for this preference, were analyzed using structural topic modeling to identify key themes and their relationships with individual difference variables. The analysis revealed five primary motives: (1) talking things over casually, (2) seeking emotional connection, (3) venting as an alternative to a close confidant, (4) seeking objective advice, and (5) consulting on topics difficult to discuss with people. Moreover, higher social anxiety was significantly associated with greater endorsement of consulting on topics difficult to discuss with people. This research provides an empirically derived taxonomy of motives for sharing emotions with conversational AI, highlighting that individuals perceive AI as a partner capable of fulfilling specific socio-affective needs by leveraging its unique features, such as its nonjudgmental nature and constant availability. The findings contribute to a deeper understanding of the evolving landscape of human–AI relationships.