Spiros I. Papadimitriou, Agapi Proimou, Vasilis Stroumpakos
https://doi.org/10.60152/m14vxiw7
Abstract: Design is fundamentally a process of modification, iteration, and re-articulation. Bruno Latour’s assertion that “to design is always to redesign” gains renewed relevance in the context of contemporary architectural practice shaped by artificial intelligence. Rather than generating wholly novel forms, AI recombines existing architectural data—typologies, spatial patterns, and materials—reinforcing Latour’s view of design as transformation rather than origination. This integration redefines authorship, positioning the architect as a mediator within a network of human and machine collaborators. Through the comparative presentation of student projects that investigate the spatial potential of AI-generated imagery, this research raises critical questions across the following themes: Bias of the Prompt: Image references carry compositional and geometric traits into AI outputs. Complex inputs yield complex results, while minimal, structured prompts offer greater control. Generative AI often produces unexpected and abstract outcomes that bypass conventional references, functioning as intermediaries in the design process. These are not final forms, but catalysts for subjective spatial exploration. Recode the Plethora: AI’s prolific output demands criteria for selection and refinement. The design process is intentionally delayed to reintroduce critique, analysis, and evaluation as core components of architectural reflection. Synergy: Iteration with AI challenges fixed notions of authorship, encouraging adaptive workflows that integrate generative tools with traditional digital and physical modeling techniques. Catalytic Hybridization: AI acts as a catalyst for spatial experimentation, enabling novel spatial blends through visual recombination. This expands the creative toolkit for designers, educators, and students alike. The research contributes to ongoing discourse on AI’s evolving role in architecture, emphasizing its ability to mediate between conceptual speculation and material organization through human-machine synergy.
Keywords: AI, Design, Iteration, Authorship, Architectural Education
How to cite this Paper (Harvard referencing style):
Papadimitriou, S., Proimou, A. and Stroumpakos, V. (2025) ‘CODE, DECODE, RECODE. AI-Generated Imagery and Architectural Design Education‘, in R. Bogdanović (ed.) On Architecture — Crosscutting and Fusion of Disciplines, Proceedings. Belgrade, Serbia: STRAND, pp. 123–138.
See publication On Architecture (2025) Conference Proceedings