Kexin Nie, Zhiwei Wang, Mengyao Guo, and Haowei Xiong. 2026. DisCraft: Exploring the Dis-embodiment of Bamboo Weaving in Generative AI. In Poster of the CHI Conference on Human Factors in Computing Systems (CHI EA '26). ACM, New York, NY, USA. （CCF-A）
Objective:
Investigate how generative AI transforms the aesthetic logic of traditional crafts, specifically bamboo weaving, by shifting it from an embodied, material-based practice to a disembodied, visually driven representation.
Methods:
· Conducted expert interviews with four national/provincial-level bamboo weaving inheritors to extract core craft principles, including structural logic, material constraints, and evaluation criteria.
· Designed a controlled comparative experiment with two AI generative paths: structure-aligned generation and pattern-translated generation.
· Applied both generative approaches across objects with varying levels of functional compatibility (high, medium, low) to analyze contextual effects.
· Performed quantitative evaluation (N=34) using Likert-scale questionnaires assessing aesthetic validity, real-world plausibility, and degree of disembodiment, supported by qualitative expert analysis.
Results:
· Demonstrated that structure-aligned generation better preserves material logic and real-world plausibility, while pattern-translated generation enhances visual expressiveness but increases perceived disembodiment.
· Revealed that evaluation of AI-generated crafts is strongly influenced by object context and functional constraints.
· Identified a fundamental shift in craft perception: from material-structural reasoning to visual-aesthetic interpretation under generative AI.
· The work was accepted at CHI EA 2026 (CCF-A), contributing to HCI, generative AI, and digital heritage discourse.
Contribution:
Contributed to research design, AI-based generative experiment implementation, and analysis of craft disembodiment, bridging generative AI with critical inquiry into intangible cultural heritage.
Abstract:
Generative AI is reshaping traditional crafts by transforming them from materially grounded, embodied practices into visually oriented representations. This shift raises critical questions about how craft aesthetics evolve when structural and material constraints are displaced.
This study investigates the case of bamboo weaving, a craft deeply rooted in material behavior and structural logic, to examine how it is reinterpreted through generative AI. Two generative pathways—structure-aligned generation and pattern-translated generation—are designed and applied across objects with varying functional compatibility.
Through expert interviews, AI-to-3D experiments, and quantitative evaluation, the study analyzes how different generative strategies influence perceptions of aesthetic validity, real-world plausibility, and disembodiment.
The results show that structure-aligned approaches maintain stronger alignment with traditional craft logic, while pattern-based approaches enhance visual diversity but weaken structural authenticity. These findings reveal a fundamental transformation in how craft is understood in the AI era, shifting from embodied making to visual abstraction.
This work contributes to the understanding of AI-mediated cultural production and provides design implications for future generative systems in digital heritage and interactive art.
Keywords:
Generative AI, Bamboo Weaving, Intangible Cultural Heritage (ICH), Digital Craft, Human-Computer Interaction (HCI)