Convincingness of AI-Generated Restaurant Reviews

Abstract

This study examines the perceived authenticity and trustworthiness of AI-generated versus human-authored online restaurant reviews. Using a randomized between subject choice experiment (evaluations n = 800), the study explores the degree to which AI-generated online restaurant reviews are distinguishable from human-authored reviews. Findings reveal that participants struggled to distinguish between AI-generated and human-authored reviews, with evaluation accuracy clustering around chance accuracy. Negative reviews were generally perceived as more trustworthy and authentic than positive ones, regardless of their source. Further, participants who reported high familiarity with online reviews had higher confidence in their evaluations despite performing no better than those with less experience. Overall, the study highlights the challenges tourism businesses face in managing the growing presence of AI-generated review content and highlights the need for robust detection mechanisms and new forms of social proof to maintain consumer trust. Future research should explore more diverse demographic samples, review contexts, or content types (e.g. AI-generated travel photos), and compare multiple frontier AI models to better understand their impact on consumer perception in real-world settings.