Title A Study on Architectural Design Methodology Utilizing Generative Artificial Intelligence and Anatomical Technique
Authors 채형주(Chae, Hyung Ju) ; 고성학(Ko, Sung Hak)
DOI https://doi.org/10.5659/JAIK.2026.42.1.139
Page pp.139-150
ISSN 2733-6247
Keywords Generative Artificial Intelligence; AI-Driven Design Methodology; ChatGPT; Stable Diffusion; Feedback Loop
Abstract This study explores the process by which artificial intelligence (AI) is integrated into architectural design, highlighting its potential to improve efficiency, creativity, and adaptability. Using Chicago's skyscrapers as a case study, we digitize architectural data and transform key properties into binary forms to create modular and scalable 3D forms. This work uses advanced generative AI tools such as ChatGPT and Stable Diffusion to develop innovative design approaches that combine algorithmic precision and creative flexibility by connecting text input with visual output. The methodology emphasizes the decomposition and reconstruction of data to incorporate the catalytic formation of Ali Rahim to match architectural forms with contextual and functional needs. The iterative feedback loop is implemented by Grasshopper, which ensures that the design output is optimized and data-driven while minimizing process conflict. Stable Diffusion provides a variety of design alternatives that enhance early-stage creativity by creating reference images. While demonstrating the innovative potential of AI, the limitations of 3D modeling precision and articulation of building elements require human intervention to refine outputs and adapt them to real-world situations. This study demonstrates AI's capacity as a collaborative partner, allowing architects to explore new possibilities while maintaining control over design intentions. Defending a balanced approach, this study highlights the complementary role of AI in restructuring architectural practices and fostering innovative and responsive design processes by combining data-driven methodologies and adaptive strategies.