Training Mode

As part of a collateral event of the Venice Architecture Biennale organized by New York Institute of Technology, faculty member Pablo Lorenzo-Eiroa led a discussion about his recently published book Digital Signifiers in an Architecture of Information (Routledge). The conversation centered around the role of big data in enabling AI models and the inherent bias within those models. Lorenzo-Eiroa questioned the currently prevalent mode of working with externally created “black box” AI models and challenged architects to get their hands dirty. “If you don’t own the means of production,” he said, “you are not the author. We have to access the processes. Otherwise we become the product.” He offered as a case study work realized in conjunction with his students, who collected immense amounts of data in the form of 3D point clouds and trained their own algorithms to imagine new neighborhoods, hybrids of existing landmarks, and altogether new forms in the urban grid. When respondent Alexis Meier raised questions about the conspicuously absent concepts of materiality and subj…