Workshop

Exploring AI’s Interpretation of the City with “La Máquina que Ve”. Workshop at Madrid School of Architecture. Conference “Artificial Architectures”

On November 14,  Iacopo Neri and Darío Negueruela led a workshop at the Madrid School of Architecture, Universidad Politécnica de Madrid, titled “La Máquina que Ve” . Through a series of hands-on exercises focused on Madrid’s architecture, we delved into how AI models interpret and categorize visual and conceptual information to make sense of our world.

This workshop highlighted the critical importance of studying machine learning models—not only because of their substantial social impact, from perpetuating biases to their disruptive potential, but also for their applicability across diverse disciplines. This is especially relevant in visually oriented fields like architecture, where AI’s role in interpretation and representation continues to grow.

Participants engaged with custom scripts designed to dissect both the training set behind CLIP and its performance when exposed to new images, seemingly unfamiliar to the model. This dual approach encouraged both a critique and appreciation of CLIP’s creative potential, allowing us to examine the concept of “Madrid” from the model’s “mind” (the data it’s trained on) and through its “eyes” (using crowd-sourced urban images).

Rodrigo Delso, Ph.D. and Enrique Villamuelas made an excellent contribution showing us their critical exploration of latent spaces andthe hacking of urban cctv systems, foregrounding the need for a creative and critical enagement with these technologies and their potential.

Conference

Artificial Architectures (AA) is the first edition of a series of conferences, round tables and workshops on new Artificial Intelligence technologies applied to Architecture to be held on 13, 14 and 15 November 2024. The series is organised by the Department of Architectural Projects of the Escuela Técnica Superior de Arquitectura de Madrid together with researchers from the Center for Digital Visual Studies DVS  (University of Zurich / Max PLanck Society), ALICE (EPFL), PROTO (UPC), Hypermedia (UPM) and the Colegio Oficial de Arquitectos de Madrid.

The increasing use of artificial intelligence in architecture, from the use of neural networks and generative models, process automation or large-scale language models, is radically transforming the way our discipline is thought and formulated. These new tools and the possibilities they open up distort an architectural education historically founded on the development of a critical and instrumental project-oriented culture, simultaneously generating paralysis, terror and fascination among members of the academic community.

To maintain our capacity to understand and act on the design of our habitats, it is urgent to attend to these transformations in a calm, informed and proactive way. To this end, the first edition of these meetings, entitled ‘Dissecting new generative technologies beyond paralysis and fascination’, will deepen our knowledge, debate and experimentation with neural networks with a view to tackling the challenges they offer us in a transversal way.

This first edition is made up of 3 workshops-conferences and 2 round tables to be held at both the ETSAM and the COAM.