AR-401(k) / 12 crédits

Enseignant: Huang Jeffrey

Langue: Anglais

Withdrawal: It is not allowed to withdraw from this subject after the registration deadline.

Remark: Inscription faite par la section


Summary

The studio examines the effects of artificial intelligence on architecture and cities. Generative tools are approached as cultural and political instruments, shaping design through data grounded in territory, economy, identity, imagery, and ecology.

Content

In the wake of rogue LLMs, autonomous agents, opaque algorithms, and AI slop, architectural tools are undergoing a profound reconfiguration. Generative design workflows and computational surrogates offer immense modeling power but also contribute to homogenization, "slopification," and erosion of critical authorship.

This studio investigates how AI might "design for itself," aiming to reclaim agency through the design of data and computational infrastructure. Rather than treating AI as a self-effacing technical container, we approach it as a social, civic, and ethical instrument, embedded in territorial, ecological, and architectural contexts.

We examine the Urban Transformer (UT) as a novel typology: a possible architecture for responsible and integrated AI infrastructure, and an inevitable one, whether we like it or not. These structures may take the form of transparent data farms, distributed urban brains, or sentient infrastructural nodes. They can appear as cohesive or decentralized systems, embedded within existing urban and suburban fabrics or as isolated techno-enclaves. The Urban Transformer becomes a test case for how architecture can shape, and be shaped by, the infrastructural needs of intelligence, balancing physical constraints with civic, ecological, and symbolic demands.

The goals of the studio are threefold: (1) to critically explore the use of generative AI in architectural design, tracing its risks and potentials beyond aesthetic automation; (2) to invent an Urban Transformer prototypology that counters the tendency toward isolated, opaque, extractive techno-capitalist systems; (3) to articulate a new architectural process and language for intelligence where form is shaped as much by energy demands, cooling logic, and physical storage as by the need for commons, symbolic resonance, public trust, and data transparency.

As AI's societal role remains contested, this studio positions architectural design as a crucial arena for negotiating how flows of intelligence become spatial, and how such architectures can mediate between technology and humanity.

This intensive studio will employ advanced digital tools and generative AI for "vibecoding" architecture. Experimental and remote LLMs and MCP agents will be introduced as exploratory digital modeling tools. A range of software, scripts, and plugins for mapping and open geodata analysis (e.g., Rhino & Grasshopper, QGIS), along with GenAI tools for representation (such as Stable Diffusion, Midjourney, and DALL-E), may act as co-design agents throughout the studio's successive phases.

No prior programming or software knowledge is required, but curiosity and strong motivation to learn are essential.

 

Keywords

  • Architectural form
  • New Typologies
  • Data-driven design
  • Artificial intelligence
  • Vibecoding
  • Urban Design

 

Learning Outcomes

By the end of the course, the student must be able to:

  • Interpret the morphogenetic parameters and other issues of relevance to the project using drawings and diagrams.
  • Critique a specific project brief and a specific context and respond with a meaningful AI-driven design concept.
  • Translate an AI-driven design concept into meaningful architectural propositions at appropriate scales and levels of granularity
  • Produce coherent architectural representations and models at sufficient levels of detail.
  • Formulate the morphogenetic narrative and create convincing arguments for the design propositions.
  • Develop convincing final diagrams, drawings, renderings, simulations, physical and digital models.

Transversal skills

  • Design and present a poster.
  • Collect data.
  • Set objectives and design an action plan to reach those objectives.
  • Make an oral presentation.

Teaching methods

  • Presentations
  • Mapping exercises
  • Hands-on design activities
  • Design reviews
  • Individual and/or Group projects

Assessment methods

Projects will be reviewed and assessed based on:

(1) their conceptual strength and innovation,

(2) the coherence and resolution of their architectural translation,

(3) their representative clarity and expressive power, and

(4) the persuasiveness of their communication, both orally, and through the physical and digital artifacts.

Resources

Virtual desktop infrastructure (VDI)

No

Bibliography

  1. Huang, Jeffrey, Mikhael Johanes, Frederick Chando Kim, Christina Doumpioti, and Georg-Christoph Holz. On GANs, NLP and Architecture: Combining Human and Machine Intelligences for the Generation and Evaluation of Meaningful Designs. Technology | Architecture + Design 5, no. 2 (2021): 207-224. https://doi.org/10.1080/24751448.2021.1967060.
  2. Huang, Jeffrey, Dieter Dietz, Laura Trazic, and Korinna Zinovia Weber, eds. Transcalar Prospects in Climate Crisis: Architectural Research in re/Action. Zurich: Lars Müller Publishers, 2024.
  3. Bender, Emily M., Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT '21),  610-623. New York: ACM, 2021. https://doi.org/10.1145/3442188.3445922.
  4. Carpo, Mario. Beyond Digital: Design and Automation at the End of Modernity. Cambridge, MA: MIT Press, 2023.
  5. Crawford, Kate. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. New Haven: Yale University Press, 2021.
  6. Del Campo, Matias, Stefan Manninger, and Andre Carlson. Archiitectural Hallucinations: Diffusion Models and New Aesthetics. International Journal of Architectural Computing 20, no. 4 (2022): 493-509. https://doi.org/10.1177/14780771221135036.
  7. Goodfellow, Ian, Jean Pouget-Abadie, Mehdi Mirza, et al. Generative Adversarial Nets. In Advances in Neural Information Processing Systems 27 (NeurIPS 2014). https://papers.nips.cc/paper/2014/hash/5ca3e9b122f61f8f06494c97b1afccf3-Abstract.html.
  8. Mattern, Shannon. A City Is Not a Computer: Other Urban Intelligences. Princeton, NJ: Princeton University Press, 2021.
  9. Zuboff, Shoshana. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. New York: PublicAffairs, 2019.

Websites

Moodle Link

Dans les plans d'études

  • Semestre: Automne
  • Forme de l'examen: Pendant le semestre (session d'hiver)
  • Matière examinée: Théorie et critique du projet MA1 (Huang)
  • Cours: 2 Heure(s) hebdo x 14 semaines
  • Projet: 4 Heure(s) hebdo x 14 semaines
  • Type: obligatoire
  • Semestre: Automne
  • Forme de l'examen: Pendant le semestre (session d'hiver)
  • Matière examinée: Théorie et critique du projet MA1 (Huang)
  • Cours: 2 Heure(s) hebdo x 14 semaines
  • Projet: 4 Heure(s) hebdo x 14 semaines
  • Type: obligatoire
  • Semestre: Automne
  • Forme de l'examen: Pendant le semestre (session d'hiver)
  • Matière examinée: Théorie et critique du projet MA1 (Huang)
  • Cours: 2 Heure(s) hebdo x 14 semaines
  • Projet: 4 Heure(s) hebdo x 14 semaines
  • Type: optionnel

Semaine de référence

Lundi, 8h - 10h: Cours

Lundi, 10h - 12h: Projet, labo, autre

Lundi, 13h - 18h: Projet, labo, autre

Mardi, 8h - 12h: Projet, labo, autre

Mardi, 15h - 18h: Projet, labo, autre

Cours connexes

Résultats de graphsearch.epfl.ch.