Paper Type
Complete
Paper Number
PACIS2026-1497
Description
The growing use of AI-generated review summaries (AIGS) represents a structural shift in how information is organized and consumed on digital platforms. Rather than simply compressing content, generative AI selectively draws from a subset of reviews to construct summaries, effectively acting as an algorithmic curator. Drawing on Accessibility–Diagnosticity Theory and Prospect Theory, this study investigates how such algorithmic selection and review intrinsic characteristics (e.g. review length, review valence) are associated with review helpfulness measured by helpful votes. Using hotel review data from TripAdvisor, we estimate a series of step wise negative binomial regression models to analyze the determinants of review helpfulness. The results show that AI-selected reviews receive significantly more helpful votes, while longer reviews and lower ratings are also associated with higher perceived helpfulness. These findings demonstrate how algorithmic curation is associated with informational value within digital review ecosystems.
Recommended Citation
LIM, SUNG JUN; Kim, Taekyung; and Lee, Dongwon, "When AI Selects Reviews: How Algorithmic Visibility Shapes Review Helpfulness" (2026). PACIS 2026 Proceedings. 6.
https://aisel.aisnet.org/pacis2026/sharing/sharing/6
When AI Selects Reviews: How Algorithmic Visibility Shapes Review Helpfulness
The growing use of AI-generated review summaries (AIGS) represents a structural shift in how information is organized and consumed on digital platforms. Rather than simply compressing content, generative AI selectively draws from a subset of reviews to construct summaries, effectively acting as an algorithmic curator. Drawing on Accessibility–Diagnosticity Theory and Prospect Theory, this study investigates how such algorithmic selection and review intrinsic characteristics (e.g. review length, review valence) are associated with review helpfulness measured by helpful votes. Using hotel review data from TripAdvisor, we estimate a series of step wise negative binomial regression models to analyze the determinants of review helpfulness. The results show that AI-selected reviews receive significantly more helpful votes, while longer reviews and lower ratings are also associated with higher perceived helpfulness. These findings demonstrate how algorithmic curation is associated with informational value within digital review ecosystems.
Comments
07-Platform