How Cross-Cue Inconsistency Weakens the Adoption of AI-Generated Review Summaries: The Roles of Diagnosticity, Authenticity, and Review Dispersion
TL;DR
The results show that AIGS-AUGR inconsistency significantly reduces consumers’ adoption intention toward AIGS, and the overall mediation analysis, perceived diagnosticity and perceived authenticity both mediate this effect.
Abstract
As AI-generated review summaries are increasingly embedded in e-commerce review systems, consumers are no longer exposed to a single type of user review information. Instead, they encounter multiple evaluative cues consisting of aggregated user-generated ratings (AUGR), original user reviews, and AI-generated summaries (AIGS). From the perspective of cross-cue inconsistency, this study examines the mechanism underlying consumers’ adoption of AIGS in a complex evaluative environment. This study adopts a scenario-based experimental method and simulates a product review page on an e-commerce platform. A 2 (AIGS: positive vs. negative) × 2 (AUGR: high vs. low) × 2 (review dispersion: low vs. high) between-subjects experimental design was employed. Data from 371 participants from mainland China recruited through Credamo online panel were analyzed using full factorial ANOVA and PROCESS-based mediation and moderated mediation analyses. AIGS-AUGR consistency/inconsistency was generated by combining the valence of AIGS with the level of AUGR, thereby capturing different manifestations of cross-cue consistency/inconsistency. Based on this design, consumers’ perceived diagnosticity, perceived authenticity, and adoption intention were measured, and the proposed hypotheses were tested. The results show that AIGS-AUGR inconsistency significantly reduces consumers’ adoption intention toward AIGS. In the overall mediation analysis, perceived diagnosticity and perceived authenticity both mediate this effect. Review dispersion differentially moderates the first-stage effects of inconsistency on the two mediators: high review dispersion strengthens its negative effect on perceived diagnosticity, whereas low review dispersion strengthens its negative effect on perceived authenticity. However, only the diagnosticity-based indirect effect on adoption intention varies significantly across dispersion levels, with high review dispersion strengthening this negative indirect effect; the moderated indirect effect through perceived authenticity is not significant.