LLM and Pattern Language Synthesis: A Hybrid Tool for Human-Centered Architectural Design
Wedi'i Gadw mewn:
| Cyhoeddwyd yn: | Buildings vol. 15, no. 14 (2025), p. 2400-2432 |
|---|---|
| Prif Awdur: | |
| Awduron Eraill: | |
| Cyhoeddwyd: |
MDPI AG
|
| Pynciau: | |
| Mynediad Ar-lein: | Citation/Abstract Full Text + Graphics Full Text - PDF |
| Tagiau: |
Dim Tagiau, Byddwch y cyntaf i dagio'r cofnod hwn!
|
| Crynodeb: | This paper combines Christopher Alexander’s pattern language with generative AI into a hybrid design framework. The result is a narrative synthesis that can be useful for informed project design. Advanced large language models (LLMs) enable the real-time synthesis of design patterns, making complex architectural choices accessible and comprehensible to stakeholders without specialized architectural knowledge. A lightweight, web-based tool lets project teams rapidly assemble context-specific subsets of Alexander’s 253 patterns, reducing a traditionally unwieldy 1166-page corpus to a concise, shareable list. Demonstrated through a case study of a university department building, this method results in environments that are psychologically welcoming, fostering health, productivity, and emotional well-being. LLMs translate these curated patterns into vivid experiential narratives—complete with neuroscientifically informed ornamentation. LLMs produce representative images from the verbal narrative, revealing a surprisingly traditional design that was never input as a prompt. Two separate LLMs (for cross-checking) then predict the pattern-generated design to catalyze improved productivity as compared to a standard campus building. By bridging abstract design principles and concrete human experience, this approach democratizes architectural planning grounded on Alexander’s human-centered, participatory ethos. |
|---|---|
| ISSN: | 2075-5309 |
| DOI: | 10.3390/buildings15142400 |
| Ffynhonnell: | Engineering Database |