From Architects to Constructors: How AI Undermines Architectural Thinking

This exploratory case study examines how text-to-image AI tools intersect with concept generation, the phase where students translate architectural theory into concrete decisions about space, form, structure, and function, in a 3rd year design studio. The research was conducted during Fall 2024 at a Central Asian university with 40 Architecture students developing a cultural center project. All participants received identical instruction on concept development; AI tools were neither assigned nor prohibited.

Documentation included semester-long observation, desk critiques, and analysis of midterm and final submissions.

Three modes of engagement emerged from observed working methods. In total, 38% of students independently incorporated text-to-image tools; 16% showed limited engagement; and 46% pursued research, abstraction, and iterative development without AI assistance. These groupings reflect behavioral patterns rather than assigned categories.

Students in the non-AI group demonstrated growing capacity to connect theoretical references to design decisions. Those using AI tools produced visually developed proposals, though the relationship between their stated concepts and subsequent spatial, formal, or functional decisions was less traceable within this single-semester context. The limited-engagement group showed minimal advancement.

At final review, differences were most apparent in how students explained the path from concept to design decisions. While individual trajectories varied, the association between mode of engagement and development of design reasoning was consistent across observed cases. These findings are specific to this studio context; they suggest questions about pedagogical sequencing when AI tools enter design education, but broader claims would require further research.