About

Kyle Steinfeld is an architect who works with code and lives in Oakland. Through his unique hybrid practice of creative work, scholarly research, and software development, he seeks to reveal certain overlooked capacities of computational design; he finds no disharmony between the rational and whimsical, the analytical and uncanny, the lucid and bizarre. His work cuts across media, and is expressed through a combination of visual, formal, and spatial material. Across these, we find an undermining of the authoritative and imperative voice that is so often bestowed upon the results of computational processes, and find in its place a range of alternative voices. His creative work at the intersection of Artificial Intelligence and Environmental Design has been exhibited at the NeurIPS workshop on Machine Learning for Creativity and design in 2017 and 2018, and has been published in Towards Data Science. In his academic and scholarly work, he seeks to illuminate the dynamic relationship between the creative practice of design and computational design methods, thereby enabling a more inventive, informed, responsive, and responsible practice of architecture. He is the author of a number of works of software design tools, including the book "Geometric Computation: Foundations for Design," which demystifies computational geometry for architecture students and design professionals. Kyle has received several research grants and fellowships, including being an IDEA fellow at Autodesk and a Hellman Fellow. As an educator, Kyle teaches core courses in design and architectural representation, as well as seminars in design computation at both graduate and undergraduate levels. His professional background includes working with and consulting for notable design firms such as Skidmore Owings and Merrill, Acconci Studio, Kohn Petersen Fox Associates, Diller Scofidio Renfro, and TEN Arquitectos. He holds a Masters of Architecture from MIT and a Bachelor's Degree in Design from the University of Florida.

Research topics

• Computer Science
• Sociology
• Artificial Intelligence
• Visual arts
• Psychology
• History
• Art
• Epistemology

Selected publications

• Intelligent tools on the loose: Reasoning models for exploratory computational design

  International Journal of Architectural Computing · 2025-08-08 · 2 citations

  articleOpen accessSenior author

  Generative artificial intelligence has swept into design research and practice, with popular image generators and language model interfaces now being used across disciplines and design stages. The multimodal language models exhibit fundamentally different characteristics from earlier machine learning approaches, and present unique challenges and opportunities for designers. A key question for design research arises: How can the capabilities of these large multimodal models become part of our design tools? This work investigates how the recent development of reasoning language models in particular can be utilized for early stage exploratory design. We outline a set of principles to build effective tools with these models by drawing from research in artificial intelligence. Specifically, we explore the application of the new reasoning model paradigm in computational design scripting. We present a system which successfully generates computer-aided design (CAD) scripts in Grasshopper. The system is compared to existing CAD AI approaches for its capabilities and characteristics. High latency and the absence of a user-model feedback loop still restrict real-time exploration in our system. Our developed principles can further inform broad exploratory tools and motivate inquiries into multimodal reasoning for design. By demonstrating a new CAD approach, we show large models’ utility as creative tools. Future research should address latency, enhance interactivity, and address other design tasks beyond CAD.

  PublisherOA PDFDOI

• Allies in Exile

  2024-02-09

  other1st authorCorresponding
  PublisherDOI

• Mediating Modes of Thought: LLM's for design scripting

  arXiv (Cornell University) · 2024-11-20 · 1 citations

  preprintOpen accessSenior author

  Architects adopt visual scripting and parametric design tools to explore more expansive design spaces (Coates, 2010), refine their thinking about the geometric logic of their design (Woodbury, 2010), and overcome conventional software limitations (Burry, 2011). Despite two decades of effort to make design scripting more accessible, a disconnect between a designer's free ways of thinking and the rigidity of algorithms remains (Burry, 2011). Recent developments in Large Language Models (LLMs) suggest this might soon change, as LLMs encode a general understanding of human context and exhibit the capacity to produce geometric logic. This project speculates that if LLMs can effectively mediate between user intent and algorithms, they become a powerful tool to make scripting in design more widespread and fun. We explore if such systems can interpret natural language prompts to assemble geometric operations relevant to computational design scripting. In the system, multiple layers of LLM agents are configured with specific context to infer the user intent and construct a sequential logic. Given a user's high-level text prompt, a geometric description is created, distilled into a sequence of logic operations, and mapped to software-specific commands. The completed script is constructed in the user's visual programming interface. The system succeeds in generating complete visual scripts up to a certain complexity but fails beyond this complexity threshold. It shows how LLMs can make design scripting much more aligned with human creativity and thought. Future research should explore conversational interactions, expand to multimodal inputs and outputs, and assess the performance of these tools.

  PublisherOA PDFDOI

• Machine Hands on Flaws to Machine: The Surprising Sources of Biases in Machine Learning Models

  Architectural Design · 2024-05-01 · 2 citations

  article1st authorCorresponding

  Abstract After musing on the history and varying media of the concept of ‘gone viral’, Associate Professor of Architecture at the University of California, Berkeley, Kyle Steinfeld further investigates computational design through the lens of cultural practices. Even the seemingly most contemporary and innovative technological ideas and gizmos can be traced back to a series of legacy notions that remain silently present in new advances. The article discusses such ‘hinge’ moments and searches for them in AI.

  PublisherDOI

• Clever little tricks: A socio-technical history of text-to-image generative models

  International Journal of Architectural Computing · 2023 · 17 citations

  1st authorCorresponding
  • Sociology
  • Computer Science
  • Artificial Intelligence

  The emergence of text-to-image generative models (e.g., Midjourney, DALL-E 2, Stable Diffusion) in the summer of 2022 impacted architectural visual culture suddenly, severely, and seemingly out of nowhere. To contextualize this phenomenon, this text offers a socio-technical history of text-to-image generative systems. Three moments in time, or “scenes,” are presented here: the first at the advent of AI in the middle of the last century; the second at the “reawakening” of a specific approach to machine learning at the turn of this century; the third that documents a rapid sequence of innovations, dubbed “clever little tricks,” that occurred across just 18 months. This final scene is the crux, and represents the first formal documentation of the recent history of a specific set of informal innovations. These innovations were produced by non-affiliated researchers and communities of creative contributors, and directly led to the technologies that so compellingly captured the architectural imagination in the summer of 2022. Across these scenes, we examine the technologies, application domains, infrastructures, social contexts, and practices that drive technical research and shape creative practice in this space.

  PublisherOA PDFDOI

• Artificiale Rilievo GAN-Generated Architectural Sculptural Relief

  Springer eBooks · 2022 · 1 citations

  1st authorCorresponding
  • Computer Science
  • Computer Science
  PublisherDOI

• Imaging Place Using Generative Adversarial Networks (GAN Loci)

  2022-05-21 · 3 citations

  other1st authorCorresponding
  PublisherDOI

• Significant others

  Routledge eBooks · 2021 · 6 citations

  1st authorCorresponding
  • Psychology

  Recent advances in the application of AI in design press us to reassess what it means to act as a creative author, and to consider new models of subjectivity that differently situate us in relation to our work. We enumerate here a number of such models that may be observed based on emerging trends in machine-augmented design practice. In summary, these are as follows: ML as actor, wherein a machine “partner” is regarded as a full-fledged actor on a par with a human designer; ML as material, wherein a generative statistical model is regarded as a new form of design “material” that is manipulated by an author in a manner analogous to certain craft traditions; and ML as provocateur, wherein AI is valued for its capacity as an instrument of the associative and imaginary. These models are illustrated by examples from current practice and suggest not only a coming shift in the way design tools operate, but that this shift might accompany a broader reconsideration of the relationship between the tools of design and the culture of architectural production.

  PublisherDOI

• Drawn, Together

  ACADIA quarterly · 2020-01-01

  articleOpen access1st authorCorresponding

  Changes in the media through which design proceeds are often associated with the emergence of novel design practices and new subjectivities. While the dynamic between design tools and design practices is complex and nondeterministic, there are moments when rapid development in one of these areas catalyzes changes in the other. The nascent integration of machine learning (ML) processes into computer-aided design suggests that we are in just such a moment. It is in this context that an undergraduate research studio was conducted at UC Berkeley in the spring of 2020. By introducing novice students to a set of experimental tools (Steinfeld 2020) and processes based on ML techniques, this studio seeks to uncover those original practices or new subjectivities that might thereby arise. We describe here a series of small design projects that examine the applicability of such tools to early-stage architectural design. Specifically, we document the integration of several conditional text-generation models and conditional image-generation models into undergraduate architectural design pedagogy, and evaluate their use as “creative provocateurs” at the start of a design. After surveying the resulting student work and documenting the studio experience, we conclude that the approach taken here suggests promising new modalities of design authorship, and we offer reflections that may serve as a useful guide for the more widespread adoption of machine-augmented design tools in architectural practice.

  PublisherOA PDFDOI

• GAN Loci

  ACADIA quarterly · 2019-01-01 · 8 citations

  article1st authorCorresponding
  PublisherDOI

Frequent coauthors

• Joy Ko

  ProHealth Care

  9 shared

• Samantha Walker

  2 shared

• Andrei Nejur

  2 shared

• Lynnette Widder

  2 shared

• Joy Ko

  2 shared

• Stefano Schiavon

  University of California, Berkeley

  2 shared

• C. Sandoval

  2 shared

• Dustin Moon

  South Dakota State University

  2 shared

Education

• Master of Architecture, Architecture

  Massachusetts Institute of Technology

  2004

• Bachelors of Design in Architecture, Architecture

  University of Florida

  1999

Awards & honors

• IDEA Studio Fellowship, "Computational Literacy for Architec…
• Faculty Research Grant, "Decod.es" (2013)
• Hellman Fellowship, "Visualization Tools for Climate-Calibra…
• Research Enabling Grant, "Dhour: A bioclimatic information d…
• Instructional Improvement Grant, UCB Office of Educational D…