AI Product Review Summaries and Content Homogenization: The Changing Landscape of UGC

Abstract

As digital platforms face growing volumes of user-generated content (UGC), many have turned to AI-generated review summaries to help users navigate overwhelming information. While these summaries aim to enhance efficiency, they may also reshape how content is consumed and produced in meaningful ways. This study examines the behavioral consequences of introducing AI-generated summaries on a major e-commerce platform. Leveraging a natural experiment on Amazon, where AI summaries were added to product pages, we combine large-scale observational data with two randomized online experiment to examine changes in both review reading and review writing behavior. We find that AI summaries reduce engagement with individual reviews. On the writing side, summaries lead to lower topical diversity, suggesting a shift toward more uniform content, potentially driven by an anchoring effect introduced by the AI-generated review summaries. The effect magnitudes vary across product types and reviewer characteristics, indicating that algorithmic mediation produces heterogeneous responses. Overall, the findings reveal that AI review summaries influence both the perceived value of reading and the expressive dynamics of review generation, reshaping the UGC landscape in the process.

Keywords: User Generated Content, Generative Models, E-Commerce Platforms, Online Reviews