From Theory to Design: AI-Supported Exploration, Critical Editing, and Student-Perceived Outcomes in Architectural Education

Generative artificial intelligence (GenAI) is increasingly used in architectural and design education, yet empirical evidence remains limited on how students use AI to translate theoretical knowledge into design decisions while retaining critical judgment. This course-embedded mixed-methods study examined student-reported and documented AI-supported theory-to-design translation in a contemporary architecture course involving ninety-six architecture students. Students selected a contemporary architectural movement and used GenAI and digital editing tools to develop a façade-screen intervention for a university building. Data included AI-use documentation, prompts, design rationales, selected movements, design goals, material and screen-element choices, written explanations, and paired student evaluations of the original and modified façade. AI use was nearly universal (95/96 students). Students rated the modified façade significantly more positively overall than the original façade (M = 4.23 vs. 3.13), t (95) = 14.02, p < .001, dz = 1.43, with higher ratings for innovation, visual comfort, light/glare/heat control, transparency/privacy balance, and inside-outside visual relation. Ratings of campus architectural identity did not increase significantly. In the exploratory regression, perceived movement suitability was the only significant predictor of modified-façade evaluation, whereas the reported number of AI alternatives was not significant. Qualitative findings indicated that students used AI to generate alternatives and translate movement characteristics into design vocabulary, while frequently requiring contextual correction and critical editing. The findings position GenAI as an exploratory support within architectural design education, where conceptual fit, student judgment, and contextual adaptation remain central to theory-to-design translation.