Discursive authenticity in AI-augmented heritage design: a critical reading of Dunhuang and Chinese embroidery

Artificial intelligence has been deeply integrated into the digital inheritance of Chinese intangible cultural heritage, covering generative models for Dunhuang murals and AI?aided design for Suzhou, Miao and Huayao embroidery. This paper argues that AI is no neutral preservation tool but redistributes discursive authenticity, the socially?negotiated authority to define authentic traditional works. Such authority is exercised not only by humans and institutions but also by technical artefacts including datasets, craft?knowledge categories and scoring metrics. Based on critical discourse analysis of 42 sources such as engineering papers, reports, policies and inheritor interviews, this study compares two cases: Dunhuang murals with few living bearers, and Chinese embroidery sustained largely by female masters. It identifies four discursive patterns, counters claims of value?neutral technology, and proposes a pluriversal ethics with seven diagnostic questions to specify requirements for legitimate cultural sovereignty.