Bridging cultural gaps in product design: A cross-cultural supervisor approach using large language model

Cross-cultural design in the globalized marketplace faces the challenges of linguistic, geographical and cultural complexity, which limits designers’ ability to create good cross-cultural product designs. This study proposes a “cross-cultural supervisor” approach based on the large language model (LLM) and the CO-STAR framework, which aims to enhance designers’ understanding of multiculturalism and optimize design concepts. By constructing a cross-cultural design metric system and quantifying cultural factors via the analytic hierarchy process (AHP), this method provides scientific support for the decision-making of cross-cultural design. We propose two main strategies: 1) optimizing cultural sensitivity and feeding back cultural knowledge in GPT-4 on the basis of the CO-STAR framework and 2) generating and evaluating design solutions via GPT-4. The effectiveness of this strategy was verified, showing that the cross-cultural supervisor approach significantly enhances cross-cultural design reports. This approach is particularly effective in design areas where cultural backgrounds and personal preferences strongly influence product outcomes. By offering insights into aesthetics, materials, and local traditions, it enables designers to adapt products to the unique cultural and behavioral expectations of diverse markets. In the era of artificial intelligence-generated content (AIGC), this study presents a new methodology for cross-cultural product design, contributing valuable theoretical and practical insights.