When and how anthropomorphic AI avatars relate to user engagement intentions: roles of emotional connection, perceived authenticity and technological familiarity

Purpose This study examines when and how anthropomorphic AI service avatars shape user engagement intentions in digital commerce. Drawing primarily on computers are social actors (CASA) theory and using parasocial interaction (PSI) as a bounded interpretive lens, it examines emotional connection as an encounter-specific affective mechanism and perceived authenticity and technological familiarity as theoretically focal contingencies operating at different stages of the engagement process. Design/methodology/approach Study 1 employed a between-subjects experiment that manipulated higher versus lower levels of bundled visual-form anthropomorphism. Study 2 used a three-wave survey of online shoppers to re-examine the focal relationships in prior AI-avatar service encounters. Mediation, moderation, and conditional indirect associations were estimated using PROCESS with 5,000 bootstrap resamples. Findings Study 1 shows that higher visual-form anthropomorphism increases user engagement intention, with emotional connection accounting for a meaningful but partial share of the effect. Study 2 provides associative support for the core relationships, although the association between anthropomorphism and emotional connection is comparatively modest. Perceived authenticity strengthens the theoretically focal association between anthropomorphism and emotional connection, while technological familiarity strengthens the association between emotional connection and engagement intention. The findings indicate that emotional connection represents one mechanism among multiple affective, cognitive and evaluative pathways linking anthropomorphic AI avatars with engagement intentions. Originality/value This study advances interactive marketing research by identifying emotional connection as a meaningful but partial encounter-level mechanism linking AI-avatar anthropomorphism with engagement intentions. It further demonstrates that perceived authenticity and technological familiarity condition the theoretically focal stages of this process, while the dominant direct component highlights the need for a multiple-pathway, rather than exclusively affective, account of AI-avatar engagement.