AI Summaries and Online Review Contributions: Effects on Modality Choice and Content Novelty

Online platforms increasingly use generative AI to summarize user-generated content and facilitate consumer decision making, yet little is known about how different forms of summaries shape users' content contribution behavior. We examine the effects of two types of AI summaries on review contributions: textbased AI summaries (TAIS), which summarize textual content only, and multimodal AI summaries (MAIS), which jointly summarize review text and images. Drawing on social comparison theory, we conceptualize AI summaries as salient comparison benchmarks that motivate contributors to differentiate their reviews from AI-generated representations of existing content and avoid redundant contributions. We propose a dual-path differentiation framework, in which contributors, motivated by a need for uniqueness, respond by (i) shifting contributions toward the modality not emphasized by AI summaries (modality differentiation) and (ii) increasing the semantic novelty of content within the modality emphasized by AI summaries (semantic differentiation). Leveraging quasi-natural experiments surrounding the introduction of TAIS and MAIS on two leading online travel agency platforms in China, we find support for this dual-path differentiation framework. Specifically, TAIS increases the proportion of multimodal reviews (i.e., reviews that combine text and images) while increasing the semantic novelty of text-only reviews. In contrast, MAIS decreases the proportion of multimodal reviews, shifting contributions toward text-only reviews, while increasing the semantic novelty of images within multimodal reviews. Together, these findings demonstrate how AI summaries affect reviewer behavior by shaping how reviewers allocate effort across modalities and differentiate from existing content. Our findings offer new principles for digital platforms to understand and manage user-generated content in the presence of AI summaries.