Paper for the International Symposium
‘Weird Science, Isolated Intelligence,’
University of Pisa, Italy, 22 February 2026.
‘Weird Science, Isolated Intelligence,’
University of Pisa, Italy, 22 February 2026.
A 1988 article published in Building Research and Practice, a journal of the International Council for Building Research (Conseil International du Bâtiment, CIB), addressed a topic more than familiar today—artificial intelligence in construction. The publication, however, steered clear of the flashy images associated with AI application today. Instead, its author, Bjørn Bindslev, a Danish architect by training and self-described “natural philosopher by inclination,” was preoccupied with how to describe architecture through a universal code. Bindlsev was not grappling with this question alone: he was part of a broader group of European building practitioners who, since the 1950s, had worked to develop a universal building classification system, a specifically architectural extension of the UDC. The system decomposed all building processes into “activities” and “resources,” each assigned a specific code. Developed in parallel with the concept of a shared European economic space, this system aimed to unify the European architectural market and establish a shared information database, connected via a network of electronic processing centres. Different versions of the system were adopted across hundreds of building projects from Scotland to Spain, while Bindslev’s own iteration, CBC (Coordination by Computer), was applied in Zambia, Kenya, and the Middle East.
Part of a broader investigation into the role of language in architecture and the problems of cultural translation, this contribution is particularly interested in the frictions between the aspired universality of this artificial system and the implicit, largely invisible codes embedded in any construction practice. Equipped with the original archives of the system’s application and the intellectual ruminations of the architects engaged with it, this contribution addresses the overlooked proto-histories of computing in architecture, before architecture became a matter of visual display. In doing so, it highlights an entirely new genealogy of textual descriptors in architecture, at the foundation of contemporary AI text-to-image generation.