What if virtual influencer’s empathic anthropomorphism is creepy? The moderating role of positive Danmu

Purpose AI-advanced virtual influencers’ (VIs’) enhanced anthropomorphism delivers substantial value yet also induces uncanny valley concerns, while relevant mitigation strategies remain under-explored. Based on the uncanny valley effect, this study explored how VIs’ empathic anthropomorphism influences user engagement intentions and verified whether social media users’ positive Danmu can mitigate the uncanny valley effect triggered by VIs’ empathic anthropomorphism. Design/methodology/approach This study employed three experiments to test the hypotheses. Experiment 1 employed a single-factor, 2-level (a VI’s empathic anthropomorphism: High vs. Low) inter-group experiment to test the hypotheses H1 to H4; Experiment 2 employed a 2 (a VI’s empathic anthropomorphism: High vs. Low) * 2 (Positive Danmu: With vs. Without) between-subjects experiment to test the moderating role of positive Danmu (H5). Experiment 3 employed a 2 (a VI’s empathic anthropomorphism: High vs. Low) × 2 (Positive Danmu: With vs. Without) between-subjects experimental in a marketing context to further verify the hypotheses H2 to H5. Findings The results show that VIs’ empathic anthropomorphism exerts a direct positive effect on users’ engagement intentions, while simultaneously inducing perceived eeriness that negatively impacts users’ engagement intentions and mediates the relationship. This perceived eeriness triggered by high empathic anthropomorphism further offsets the positive influence of VIs’ empathic anthropomorphism on users’ behavioral intentions. Most notably, the presence of positive Danmu effectively mitigates the effect of VIs’ empathic anthropomorphism on users’ perceived eeriness. Originality/value First, it addresses a research gap in VI emotions, offering theoretical insights into human–computer interaction, focusing on empathic anthropomorphism’s impacts. Second, our work provides important insights into uncanny valley research in the VI marketing by exploring how VIs’ empathic anthropomorphism shapes users’ perceived eeriness and engagement intentions. Third, we make an important contribution to the study of boundary conditions in VI marketing research by investigating the moderating role of positive Danmu. The research findings have considerable theoretical and practical inspirational value in terms of how to overcome the uncanny valley effect while developing VI anthropomorphic technologies.