Skip to content

Author

Weifeng Hu

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#generative ai Open access Aug 2026

From aesthetics to authenticity: a stimulus–organism–response model of audience responses to AI-generated intangible cultural heritage design

Generative AI is increasingly used in heritage visualization, yet its outputs often raise concerns about cultural authenticity. Prior studies have focused more on technical fidelity and symbolic preservation than on how audiences evaluate the authenticity of AI-generated heritage imagery. To address this gap, this study develops and tests a Stimulus-Organism-Response model using AI-generated Wuxi clay figurine images. A randomized between-subjects online experiment with 330 participants examined the associations of visual information quality, AI technical novelty, and symbol salience with perceived aesthetics, perceived authenticity, cultural identity, and acceptance intention. The theory-specified model showed that visual information quality, technical novelty, and symbol salience were positively associated with perceived aesthetics, while perceived aesthetics and symbol salience were positively associated with perceived authenticity. Perceived aesthetics and perceived authenticity were also associated with cultural identity and acceptance intention. However, the HTMT analysis indicated limited discriminant validity, particularly among visual information quality, technical novelty, symbol salience, and perceived aesthetics. A supplementary second-order model representing these four perceptions as facets of an overall perceived design quality factor showed acceptable fit and comparable downstream associations. In this alternative specification, perceived authenticity was no longer independently associated with acceptance intention. Accordingly, the construct-specific path estimates should be interpreted cautiously. Overall, the findings are consistent with the presence of a substantial holistic evaluative component in audience responses to AI-generated heritage imagery.

Lu Feng, Weifeng Hu · 0 citations