Can Dominant Architectural Culture Influence Cognitive Processes? Architectural Intelligence and AI-Assisted Evaluation

The concept of technological singularity is discussed here in the context of architecture (of buildings, not software). This is the point at which non-human intelligence is conjectured to surpass ordinary human cognitive limits. Empirically constrained AI may already offer a useful corrective to mainstream architectural culture in one crucial aspect—its capacity to evaluate design that adapts to human emotional health. Postwar building architecture as an institutional power system rewards abstraction and stylistic conformity through media prestige while not always accounting for embodied human experience. By narrowing judgment criteria, architectural studio pedagogy trains tacitly for imitation, not seeking evidence that conflicts with dominant formal ideologies. Yet findings from environmental psychology, health-related design research, neuroscience, and recent AI-based studies show that built form measurably affects empathic response and user well-being. This paper examines what effects dominant architectural culture could impose on the public by producing informationally impoverished, stressful environments. We argue that built environment design may suffer from an epistemic closure because (i) architectural education does not foster curiosity in how design affects users—the core mechanism for intelligence development—and (ii) architectural media may legitimate non-adaptive form languages by habituating populations to ignore distress signals from geometries associated with elevated stress markers. However, empirically constrained AI can now be directed to apply that relevant knowledge base to improve the built environment. The most suggestive evidence in the paper is that LLM emotional scores, LLM geometric scores, human eye-tracking, and large public surveys converge on the same designs. In this sense, the AI singularity can be framed as a domain-specific, testable hypothesis in architecture. This paper does not report new generated results derived from Empirically Constrained Scaffolding (ECS), which appear in prior applications, but reproduces the original prompts as an illustration of the method.