From the Drawing Board to the Prompt: A Systemic Review of Architectural Representation Paradigms between Analogy, Digital Rendering, and Artificial Intelligence Generation

Abstract

Jorge Pablo Aguilar Zavaleta and Maria Laura Lopez Luna

Contemporary architectural representation is undergoing an accelerated epistemological transformation derived from the coexistence between analog drawing, digital rendering and generative artificial intelligence. Far from constituting a linear sequence of technological substitution, these paradigms configure differentiated regimes of cognitive production, formal control, and communicative mediation. The study critically analyzes the continuities and ruptures between these systems of representation, with the aim of identifying their technical foundations, operational scopes and limitations in contemporary architectural practice.

The research is based on the hypothesis that the effectiveness of each medium depends less on its technological sophistication than on its adaptation to specific phases of the project process and to different audience profiles. A qualitative approach of critical and comparative systematic review was adopted, based on literature indexed in Scopus, Web of Science and Google Scholar between 1990 and 2026, also incorporating seminal texts on architectural theory and visual representation. The final corpus was made up of 52 sources distributed between analog, digital and generative AI paradigms.

The analysis was developed using thematic coding in two cycles following the Braun and Clarke protocol and a multidimensional comparison matrix based on variables such as production time, geometric precision, iteration, authorial expressiveness, accessibility and control over the result. The findings show that manual drawing retains relevant cognitive advantages during the initial phases of ideation, particularly due to its ability to activate processes of divergent exploration and spatial reasoning associated with project thinking. In contrast, digital representation offers operational superiority in geometric precision, interdisciplinary coordination, and reproducibility, although it introduces phenomena of visual homogenization and early dependence on defined geometries. The most disruptive result corresponds to generative AI: while it drastically reduces viewing times and expands accessibility for users without graphic training, it also shifts authorship from direct production to curatorial processes and algorithmic selection.

Counterintuitively, recent studies cited in the article show that AI-generated images are perceived as equivalent to or superior to professional renderings by non-specialized audiences, while expert architects continue to detect spatial and constructive inconsistencies. The study concludes that contemporary architectural practice operates under a post-digital logic where no paradigm is self-sufficient. A layered integration model is proposed that articulates analogue ideation, digital development and AI-augmented variation as a strategy for balancing creativity, precision and communicative efficiency. Likewise, it is warned that the premature replacement of manual drawing in architectural training could affect cognitive abilities linked to spatial reasoning and early conceptual generation.

PDF