Human vs. AI Influencers: Understanding Consumer Engagement and Brand Attachment

Abstract

The swift rise of computer-generated “virtual” or artificial-intelligence (AI) influencers, in parallel with traditional human endorsers, has prompted brands to reconsider which type of influencer more effectively fosters consumer engagement and brand loyalty. This study utilizes source credibility theory, parasocial interaction theory and the uncanny valley framework to analyze consumer reactions to both human and AI influencers across six dimensions: perceived authenticity, source credibility, parasocial interaction, consumer engagement, brand attachment and purchase intention. A between-subjects survey involving 320 participants was conducted, exposing them to a social media endorsement campaign featuring either a human or an AI influencer. Results from independent-samples t-tests indicated that human influencers outperformed their AI counterparts across all dimensions. The most significant disparities were observed in perceived authenticity (d = 0.93) and parasocial interaction (d = 0.86), while the difference in consumer engagement was notably smaller (d = 0.24). Hierarchical regression analysis revealed that parasocial interaction and source credibility were the strongest predictors of brand attachment alongside engagement together accounting for 59.1% of variance. Interestingly, the type of influencer lost its predictive power when these psychological factors were considered. Additionally, a moderation analysis pointed to a compensatory trend: after adjusting for perceived authenticity, the engagement gap between the two types of influencers narrowed to an insignificant level. These findings lend support to a mechanism-based perspective where the type of influencer influences engagement and brand attachment indirectly via credibility and relational dynamics rather than solely through the distinction between human and AI influencers. The study concludes with managerial insights, theoretical implications, limitations and suggestions for future research avenues.