Abstract
This article examines the evolution of computational paradigms in architecture through the articulation of a diachronic framework and a comparative analytical matrix. Moving beyond linear narratives centred on technological progress, the study proposes an interpretation of architectural computation as a layered ecology in which distinct regimes—symbolic, representational, informational, generative, and probabilistic— interact simultaneously. Based on a critical review of historical, theoretical, and technical sources, the study comparatively examines five major paradigmatic moments in the development of architectural computation. Instead of proposing these paradigms as discrete or sequential stages, the article interprets them as interdependent computational layers that continue to coexist within contemporary architectural practice. The findings indicate that the transition from rule-based deterministic systems to learning-based systems introduces a fundamental shift in the nature of architectural computation, moving design processes from controlled execution toward probabilistic exploration. In this context, artificial intelligence does not merely extend existing technical capabilities but reconfigures the relationships between designer, tool, and knowledge. The article concludes that contemporary architecture operates within a layered computational ecology in which multiple paradigms overlap and interact. This perspective allows computation to be understood not only as a set of tools but as an epistemological infrastructure that profoundly transforms architectural practice, its processes, and its critical frameworks.