Answers are grounded in this paper and cite their sources.
This essay argues that the most consequential question for AI in architecture is not whether AI can generate buildings, but whether it can be guided toward designs that are coherent, beautiful, and supportive of human life. To move past the present impasse of style-driven generation, the essay develops a Living Structure + AI paradigm grounded in Christopher Alexander's theory and operationalised through a skeleton–skin workflow. The skeleton encodes hierarchical order through nested substructures and computable indicators — the L-score and the B-score produced by the Beautimeter; the skin is generated by AI under structural constraint. The paradigm is illustrated through four real renovations on the HKUST(GZ) campus and connected to recent — and still preliminary — results in which large language models, properly prompted, judge architecture in broad agreement with public preference and with measured brain responses. AI for architecture, on this account, is less about spectacle than about the quiet, demanding work of structural diagnosis and structurally constrained generation.
No takes yet. Share an insight, caveat, or question.
Bin Jiang (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: