Integrating Generative AI into the Conceptual Design Process: A Case Study of Hydrogen-Powered Water Mobility

초록

Background Generative AI is transforming art and design by enhancing aesthetics and shape grammar in ideation. However, its linear thinking limits handling ambiguous tasks without clear intent. Thus, designers’ expertise and evaluation are crucial for consistent, objective outcomes. This study examines generative AI in supporting concept development and visualization in early design stages. Methods We derived three design concept keywords—hydrogen, waterborne, and mobility—through preliminary research and, based on these, curated images generated in Midjourney. Guided by the selected images, we modeled the form, then extracted and applied CMF (color, material, finish) schemes using Vizcom to refine the model, and produced the final design via rendering. The final proposal was evaluated through expert in-depth interviews and an online survey. Results The Midjourney, Vizcom, and final design image went through a comparative evaluation on formal quality, colorfulness, feasibility, and novelty, revealed that feasibility and formal quality were rated higher in human-refined designs, while colorfulness and novelty received lower ratings. Furthermore, tangible form-related keywords such as mobility, futuristic, and ferry scored significantly higher than abstract concepts such as eco-friendly public transportation, sustainability, and hydrogen energy. Conclusion Generative AI supports form exploration and visual diversity in early ideation; however, feasibility and concept coherence rely on the designer’s critical judgment and selective refinement. To ensure meaningful outcomes, designers must properly handle the alignment of AI-generated results with the intended concept—an evaluative task beyond AI’s capability. Therefore, generative AI should serve as a supportive tool, while final design quality is ultimately shaped by the designer’s intent and interpretation.

키워드