A Human-Centered AIGC Framework for Inclusive Fashion Design: Mitigating Bias for East Asian (Chinese) Elderly

This paper applies human-centered Artificial Intelligence-Generated Content (AIGC) techniques to fashion design for the East Asian elderly population, with a specific focus on the Chinese cultural context, situated within the Industry 5.0 framework. While AIGC offers substantial opportunities for design automation and personalization, mainstream applications frequently exhibit algorithmic bias and lack cultural inclusivity. To address these challenges, we propose and evaluate a human-centered AIGC framework that integrates deep personalization—user-specific model training using Low-Rank Adaptation (LoRA)—with designer-centric evaluation via structured assessment using the Fuzzy Analytic Hierarchy Process (FAHP). Employing a mixed-methods approach, which combines two design experiments and FAHP analysis involving 22 designers and experts, we investigate how AIGC can enhance fashion design for this marginalized demographic. The first experiment examines AIGC’s potential for digital prototyping to reduce material waste, directly supporting Industry 5.0’s sustainability pillar. The second experiment utilizes LoRA technology to create a deeply personalized collection for a representative Chinese user, demonstrating a viable pathway to mitigate representational algorithmic bias and improve cultural inclusivity. FAHP analysis reveals that designers prioritize AIGC for enhancing design efficiency, iteration speed, innovation, and diversity. The findings culminate in an integrated framework that leverages AIGC for sustainable and socially responsible design, providing practical insights for the development of human-centered AI applications that empower both designers and users.