A hybrid human-AI collaborative workflow for digital heritage interpretation: a case study on the “Dongdong Push” Nuo mask

Abstract

With the development of generative artificial intelligence (AIGC), its application in human–AI collaborative design has attracted increasing attention. However, in the process of intangible cultural heritage (ICH) digital revitalisation and cultural product design, systematic research on balancing traditional cultural characteristics with contemporary market aesthetics while avoiding AI-generated cultural bias and visual homogenisation remains limited. Taking the “Dongdong Push” Nuo mask of the Dong ethnic group in China as a case study, this paper proposes an AIGC-assisted ICH IP design workflow based on human–AI collaboration. By integrating the Double Diamond design model, large language models, and AI image-generation tools, the study reconstructs the traditional ICH IP design process. Consumer preference analysis and cultural feature extraction are employed to establish semantic constraints for AI generation, while an AHP-FCE-based multi-stakeholder evaluation framework is constructed to assess design effectiveness through questionnaires and semi-structured interviews. The results indicate that the proposed workflow supports efficient trend extraction, style exploration, and visual ideation while achieving overall performance comparable to traditional manual design. The quantitative comparison revealed only small differences across evaluation dimensions, while the qualitative findings highlighted potential benefits of AIGC in visual exploration and iterative ideation. At the same time, the application of AIGC remains subject to challenges such as visual homogenization, AI hallucinations, and aesthetic inertia. These findings suggest that the value of AIGC in ICH design lies not in replacing human designers, but in supporting and expanding their cognitive and creative exploration through human–AI collaboration.