EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Beyond the label itself: how disclosed content shapes consumer responses to AI-generated product imagery in E-commerce

Xiran Yang, Jiaxi Liu, Changlin He, Junjie Chu

Frontiers in Computer Science · 21.08.2026 · 8

DOI: 10.3389/fcomp.2026.1860932 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Computer Science, online erschienen am 21.08.2026. Verfasst von Xiran Yang und 3 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

With increasing global demands for transparency, disclosure labels are becoming a standard feature of e-commerce interfaces, and a central open question is how the presence of such a label, and what it discloses, changes consumer responses relative to no label at all. This study examines this question using AI-generated content as a focal, currently topical case of disclosed label content, alongside human-generated content.

Methods

Displayed on e-commerce platforms such as Taobao, we employed a 3 × 2 between-subjects experimental design that manipulated disclosure condition (AI-generated label, human-generated label, no label) and product type (utilitarian vs. hedonic), analyzed using a conditional process (moderated-mediation) framework with bootstrapped confidence intervals.

Results

In the utilitarian product context, AI disclosure labels reduced perceived authenticity, esthetic appeal, and social presence relative to no label. Product type significantly moderated the first-stage effect of AI labels on esthetic appeal and social presence: the negative effect of the AI label on these two pathways was concentrated in the utilitarian product context and was not detectable in the hedonic context (index of moderated mediation confidence intervals excluding zero for both pathways and both downstream outcomes). Esthetic appeal and social presence, in turn, positively predicted both purchase intention and brand trust. Independent of these cognitive pathways, AI disclosure labels carried a positive direct effect on purchase intention (b = 0.308, p = 0.010) that was not moderated by product type, while showing no significant direct effect on brand trust. Because this direct effect ran opposite in sign to the negative indirect pathways through esthetic appeal and social presence, the overall (total) effect of AI labels appeared weak or even negative in raw form, a pattern consistent with inconsistent mediation (suppression).

Discussion

These findings refine the traditional source-information consistency account of algorithm aversion and highlight the need for context-sensitive, platform-level governance of AI-generated product imagery in e-commerce.

Zitieren

APA

Yang, X., Liu, J., He, C., & Chu, J. (2026). Beyond the label itself: how disclosed content shapes consumer responses to AI-generated product imagery in E-commerce. Frontiers in Computer Science, 8. https://doi.org/10.3389/fcomp.2026.1860932

Vancouver

Yang X, Liu J, He C, Chu J. Beyond the label itself: how disclosed content shapes consumer responses to AI-generated product imagery in E-commerce. Frontiers in Computer Science. 2026;8. doi:10.3389/fcomp.2026.1860932

BibTeX
@article{Yang2026,
  author = {Yang, Xiran and Liu, Jiaxi and He, Changlin and Chu, Junjie},
  title = {Beyond the label itself: how disclosed content shapes consumer responses to AI-generated product imagery in E-commerce},
  journal = {Frontiers in Computer Science},
  year = {2026},
  volume = {8},
  doi = {10.3389/fcomp.2026.1860932}
}
Kostenlos starten →