Transcending time: bridging cultural heritage and contemporary creativity in ‘ AI generated art ’

This paper examines how AI artists working with large pre-trained models selectively retrieve materials from cultural heritage as stored memory and reinterpret them in digital-art contexts. Using the framework of Cultural Memory, we argue that generative models enable the selective mobilization of heritage materials within contemporary public discourses. Combining theoretical analysis, case studies, and practice-based research, we first analyze three AI art projects to demonstrate how artists selectively retrieve materials from cultural heritage as stored memory and use them to address contemporary concerns, including cultural invisibility, linguistic authority, human–machine hierarchy, and difficulties experiencing intimacy. Through reinterpretation, stored memory becomes potential functional memory as heritage materials acquire new artistic and discursive uses. We then extend this framework through our project series Fusion: Landscape and Beyond (2022–2023), which reimagines Chinese shan shui painting via dataset curation, model fine-tuning, text-to-image generation, animation, and real-time interactive installation. Transforming urban imagery into shan shui-inspired visuals, the project highlights the ontological gap between idealized ecological aesthetics and ecological anxiety in the urbanized Anthropocene. We suggest that generative AI can assist in reinterpreting heritage materials in response to contemporary issues, inviting further exploration of technology’s role in constructing and evolving human wisdom.