Who Should Disclose AI-Generated Reviews? Evidence from Two Experiments

초록

Purpose – With the rapid diffusion of generative artificial intelligence, AI-generated content is increasingly prevalent on online review platforms, raising concerns about credibility, transparency, and consumer trust. This study examines how AI-generation disclosure and discloser identity affect consumer evaluations of online hotel reviews. Design/Methodology/Approach –­ Two controlled experiments test whether consumers can distinguish AI-generated from human-written reviews, and how disclosure strategies shape perceived information credibility, information quality, and review usefulness. Findings – Study 1 shows that without disclosure cues, most consumers cannot accurately identify AI-generated reviews; once labeled as AI-generated, perceived credibility, quality, and usefulness decline significantly. Study 2 compares platform-initiated disclosure with user-initiated disclosure in which reviewers voluntarily indicate AI generation. User-initiated disclosure yields higher perceived credibility, quality, and usefulness than platform disclosure. Mediation tests indicate that the advantage of user-initiated disclosure operates through increases in perceived information credibility and information quality. Trust propensity further moderates these effects; consumers with lower trust propensity respond more strongly to disclosure cues, whereas those with higher trust propensity respond less. Overall, the findings suggest that while disclosure supports transparency, the identity of the discloser critically shapes consumer trust in AI-generated reviews. Research Implications – The findings suggest that platforms should move beyond compliance-driven disclosure toward human-centered transparency designs that empower voluntary labeling and account for individual differences in trust propensity.

키워드