Generative-inferential spatial-semantic AI: normative rational social imperatives at another socio-technical horizon.

Environment scanning technologies combine sensing methods like light detection and ranging (LiDAR), photogrammetry, and simultaneous localization and measurement (SLAM) to describe existing architecture using point cloud and mesh geometry. Semantic segmentation, the process of classifying point cloud geometry into categories of real-world objects, sees increased automation via large-scale computing models, or artificial intelligence (AI). It is feasible that, analogous to image-generation models like Midjourney, semantic segmentation could be reverse-engineered to generate coherent point cloud geometry and new, digital architecture from existing data. This contribution will review emerging, possible technological progressions to support a position on its relationship to architecture and social change. It will argue that, viewed considering critical sociological theory, the technical horizon of a generative-inferential spatial-semantic (GISS) AI should prompt architects to develop new technical competencies, but also to re-centralize normative social imperatives in practice and pedagogy. Following from critical sociological theorists Theodor Adorno and Max Horkheimer, and globalization theorists Leslie Sklair and David Held, the paper will contrast normatively and technically rational responses to technological change.