Title A Comparative Analysis of Creative Methods in Architectural Design Using Generative AI and Sketch-Based Approaches
Authors 강승현(Kang, Seoung-Hyun) ; 김덕수(Kim, Duksu)
DOI https://doi.org/10.5659/JAIK.2026.42.6.75
Page pp.75-86
ISSN 2733-6247
Keywords Diffusion-Based Generative AI; Traditional Design Tools; Creativity; Midjourney; Architectural Drawing; Sketch; Generation
Abstract This study examines differences in creativity in architectural design between an AI-based image generation tool Midjourney and traditional hand sketching. As artificial intelligence becomes increasingly integrated into design practice, it is important to determine whether these tools genuinely enhance creative potential or simply provide faster visual output. The research compares AI-based and traditional methods in generating design solutions, focusing on originality and contextual appropriateness. An experiment was conducted using a single-family house design task, producing both AI-generated and hand-sketched outcomes. These results were evaluated through participant surveys and expert assessments based on several creativity-related criteria: interestingness, novelty, appropriateness, thoughtfulness, immediacy, visual impact, and practicality. The findings show that Midjourney excels at rapidly producing visually striking and diverse design alternatives, particularly in terms of interestingness, novelty, and speed. In contrast, hand sketching is more effective in addressing programmatic requirements, spatial coherence, and contextual fit, resulting in higher evaluations for appropriateness, practicality, and thoughtful integration of constraints. Overall, AI-based tools are best understood not as replacements for traditional methods, but as complementary tools that enhance early-stage design exploration. Future research should further develop hybrid workflows that strategically combine AI-generated imagery with sketch-based refinement to improve both efficiency and depth in the architectural design process.