The proliferation of AI-generated content (AIGC) inevitably surrounds consumers with mixed-source content without clear authorship disclosure, prompting questions about how they perceive and respond to AI-like content. This study aims to examine how consumers infer AI involvement and how such subjective perceptions affect their content evaluation.
With a focus on online review settings, we propose that features in the sentiment (sentiment wave), syntactic (descriptive and formal writing styles) and pragmatic (redundant content) layers of reviews evoke consumer suspicion of AI involvement in review writing. We argue that perception of AI authorship diminishes evaluations of review helpfulness, primarily through eroding perceived authenticity. Three complementary studies validate our assertions.
Results revealed limited consumer ability to accurately detect AI authorship. The negative association between the perception of AI authorship and review helpfulness was stronger for experience (vs. search) products, and enhancing AI interpretability failed to mitigate consumers’ aversion to AI-like online reviews.
The findings guide content creators to avoid unwarranted AI-like signals when crafting authentic reviews, inform AI literacy programs aimed at improving consumers’ detection accuracy and content evaluation and help platforms design context-specific disclosure policies that safeguard authentic content.
These results provide novel insights into subtle consumer content adoption behaviors in the era of generative AI, advancing understanding of algorithm aversion in anonymous, suspicion-driven environments.
