Paper Number

ICIS2025-2441

Paper Type

Short

Abstract

AI Summaries with Images (ASI) is an emerging phenomenon in online reviews, leading to a Janus effect of human-AI interaction. Our study suggests that ASI may alter existing theoretical findings on AI summaries. Using a unique panel dataset of matched hotels on Qunar.com and Ctrip.com, we exploit Qunar’s quasi-experimental introduction of ASI to examine its impact on online reviews. Our results reveal surprising findings: the adoption of ASI unexpectedly reduces the proportion of image reviews while increasing both the proportion of text-only reviews and the number of reviews. The shifts in review proportions are primarily concentrated among higher-rated hotels, whereas the effect on the number of reviews does not vary with hotels’ rating levels. To the best of our knowledge, this study provides the first empirical exploration of ASI and gives a more nuanced understanding of human-AI interaction patterns, thus contributing new theoretical insights.

Comments

19-SharingEconomy

Download

Share

COinS
 
Dec 14th, 12:00 AM

Examining the Janus Effect of Human-AI Interaction in AI Summaries with Images

AI Summaries with Images (ASI) is an emerging phenomenon in online reviews, leading to a Janus effect of human-AI interaction. Our study suggests that ASI may alter existing theoretical findings on AI summaries. Using a unique panel dataset of matched hotels on Qunar.com and Ctrip.com, we exploit Qunar’s quasi-experimental introduction of ASI to examine its impact on online reviews. Our results reveal surprising findings: the adoption of ASI unexpectedly reduces the proportion of image reviews while increasing both the proportion of text-only reviews and the number of reviews. The shifts in review proportions are primarily concentrated among higher-rated hotels, whereas the effect on the number of reviews does not vary with hotels’ rating levels. To the best of our knowledge, this study provides the first empirical exploration of ASI and gives a more nuanced understanding of human-AI interaction patterns, thus contributing new theoretical insights.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.