Mechanical, Thinking, or Feeling? The Differential Effects of AI-Generated Reviews on Consumer Trust
Yizhen Wei, David (Jingjun) Xu, and Kai Li
International Journal of Electronic Commerce,
Volume 30, Number 4, 2026, pp. 567-598.
Abstract:
Existing research shows that potential consumers rely heavily on consumer reviews (i.e. consumer-generated reviews) when purchasing. However, many products lack sufficient consumer reviews. With advances in AI technology, platforms have begun using AI-generated reviews, generated using deep learning based on reviews of the same product from other sellers and platform databases. Such reviews are intended to serve as substitutes when consumer reviews are unavailable. However, the efficacy of AI-generated reviews remains unclear, particularly in building consumers’ trust as effectively as genuine consumer reviews. Drawing on the service literature, this study classifies AI-generated reviews into three types, namely, mechanical, thinking, and feeling. Then, it examines their effects on consumer trust through a laboratory experiment. The results reveal that different types of AI-generated reviews vary substantially in their trust-building. Specifically, consumers trust mechanical AI-generated reviews less than consumer reviews, because mechanical AI-generated reviews are perceived as less grounded in authentic experience. Among the three AI-generated review types, thinking AI-generated reviews are the most effective in enhancing consumer trust and emerge as the most promising substitute when consumer reviews are unavailable. When consumer reviews are present, feeling AI-generated reviews become a more viable option. These findings advance the theoretical understanding of AI-generated content and offer practical guidance for sellers and platforms in selecting the most appropriate AI-generated review type based on consumer review availability.