Every Dataset is a Canon

For architects, the construction of a dataset for use in AI, whether imported from somewhere else or newly created, is fundamentally an epistemological canon of references, likes and preoccupations often predicated on the selection of so‐called masterpieces of Western art. The canon represents ideals of beauty and knowledge, but it also reflects a bias in favour of art created by those who have occupied the most socially, politically and economically powerful positions in culture. Mario Carpo, architectural historian, critic and Reyner Banham Professor of Architectural History and Theory at the Bartlett School of Architecture, University College London, leads us through a succinct series of eras and architectural styles to demonstrate this fact and their lack of diversity.