AI (Artificial Intelligence) and the Reconfiguration of Architectural Practice
- mborsett
- 3 hours ago
- 7 min read
Written by Charles Fang @MBA architects
Artificial Intelligence (AI) has moved rapidly from a peripheral technology to a structuring force across the architecture, engineering, and construction (AEC) sector. It promises increased efficiency, new modes of creative exploration, and unprecedented data integration, while simultaneously raising profound questions about authorship, ethics, and the social value of architectural labor. Drawing on a broad overview of AI as well as recent industry and educational reports, we examine how AI is transforming architectural practice and pedagogy. Together, these sources reveal a field characterized by accelerated adoption, uneven strategic integration, and a re-centering of what it means to teach and practice architecture in the age of intelligent machines.
The (Historical) Promise of Computation
Artificial intelligence (AI) is not a sudden rupture in architectural history but an intensification of a century-long convergence between design, technology, and algorithm. Sean Keller’s essay on the Seagram Building traces architecture’s computational turn to Lionel March’s Cambridge laboratory in the 1970s. March envisioned making architecture more rational and scientific, but instead succeeded in creating a new architectural vocabulary that were “[in digital environments] freed from the constraints of actual buildings.”
The evolution of simple building forms over time, gradually transforming from basic structures into more complex configurations, visually illustrating architectural development. (Gif credit: WIX)
Antoine Picon, in “The Seduction of Innovative Geometries,” argues the computer allowed architects to think of form not as a fixed geometry but as a process, marking the beginning of architecture’s algorithmic imagination. Yet, he cautions, it also risks becoming self-referential: an endless pursuit of novelty detached from ethics, place, and society. This critique sets the stage for understanding how contemporary AI continues—and complicates—this trajectory.
In parallel, artificial intelligence experienced a Copernican revolution when computational systems started to imitate the human nervous system and the advance in processing capabiility allowed deep learning, the core of the AI we know today. In “Machine Learning: The New AI”, Ethem Alpaydin describes how early artificial intelligence began with rule-based reasoning. By the late twentieth century, the field underwent a paradigm shift toward data-driven learning—systems that inferred patterns rather than obeyed rules. The 2016 victory of the AlphaGo program by Google’s Deepmind over the game, Go’s, top-ranked player is a landmark in this new direction of machine learning . The subsequent explosion of the large language model, Chatgpt, in 2022 leads to our current landscape of text and image generators.
Practical Utility and Uneven Adoption
Of the recent industry reports on AI’s integration into professional practice, one consistent theme emerges: the fastest and most sustained adoption of AI in architecture has been in textual and administrative workflows rather than in image generation or form-finding. The Architects’ Journal (2025) survey found that approximately 64% of respondents used text-based AI tools weekly, with 77% using ChatGPT specifically—far higher rates than for image generators such as Midjourney or DALL-E. Architects primarily applied AI to routine yet time-intensive tasks such as bids, planning statements, reports, and fee proposals.
By contrast, AI image-generation tools remain viewed with skepticism. Many architects perceive the results as generic, stylistically homogenous, or unsuitable for high-stakes presentation work. The Deltek YouTube discussion cautions that while AI can accelerate ideation, it cannot yet replace the aesthetic and contextual judgment of human designers. In short, AI has proven its worth in augmenting communication and production, but less so in authentic design authorship—a distinction that underscores architecture’s dual identity as both business and art.
From “Shadow AI” to Strategic Integration
While adoption is widespread, the infrastructure to manage it is lagging. The Architects’ Journal report describes a landscape of “grassroots experimentation in a governance vacuum”. Individuals independently experiment with AI tools, often outside formal company policy, creating a culture of shadow AI. Firms rarely have dedicated AI budgets, training programs, or clear ethical frameworks , and the individual use of AI outside sanctioned use may introduce cybersecurity risks, a primary concern of larger firms. In contrast, firms that embed AI within their broader business strategy—investing in data management, cybersecurity, and continuous learning—tend to generate more sustainable value. Deltek’s presenters argue that “you don’t need an AI strategy; you need to embed AI into your business strategy”. This insight reframes AI not as an external tool but as a structural component of how design knowledge, data, and intellectual property circulate through a firm.
Expanding Risk: Data and Accuracy
The Architects’ Journal survey adds a professional dimension: architects cite persistent accuracy issues as the largest barrier against using AI which also brings to bear questions of accountability. For example, who bears responsibility if an AI-generated calculation or code analysis is incorrect? There are also concerns about intellectual property and biases. These concerns are especially poignant as they most pointedly affect textual and administrative workflows for which firms overwhelmingly utilize AI. If inaccuracies or biases enter bids, planning statements, and reports, these may substantially harm professional relationships or fundamentally change the character of projects, ultimately to the detriment of the built environment. Without transparent data provenance and validation processes, the line between legitimate assistance and dangerous quagmire remains blurred.
AI and Architectural Education: From Product to Process
The RIBA article extends these industrial concerns into the pedagogical realm, with Professor Des Fagan responding to the Standing Conference of Schools of Architecture (SCOSA) report, “AI and the Future of Architectural Education in the UK”. It reports that students can now “generate convincing images, models, design options and essays by using just a few language prompts—all without any understanding of the logic of the process involved”. This capacity simultaneously empowers and destabilizes architectural education.
The 'Swoosh' pavilion, which won a competition, was designed and constructed by student Valeria Garcia Abarca and her team from the Architectural Association. Wikimedia Commons© 2008. Photo by Ronaldccwong
RIBA’s analysis calls for critical thinking rather than mere tool proficiency. Students must learn to interrogate how algorithms work, what data they rely on, and what biases shape their outputs. Importantly, the article reframes the value of architectural education: “Generative AI challenges us to reflect on the true nature of the work we complete as architects… AI may accelerate a shift to recognizing that an architect’s true value is not the image, but the problem solving, the iterative development, the critical thought and the human conversations that lead to it”. One can only hope that firms will further this call, seeking graduate students entering professional practice who can not only operate AI tools but also question and guide them.
Structural Tensions and Emerging Ethics
A deeper synthesis of these sources reveals structural tensions that go beyond workflow efficiency. AI increases productivity but risks compressing professional labor, potentially devaluing entry-level tasks that once served as training grounds for emerging architects. As design production becomes semi-automated, the locus of expertise may shift from drawing to data management, from individual authorship to collective curation. Those who control datasets, prompt libraries, and algorithmic integrations will likely capture disproportionate economic value. Indeed, Deltek’s Innovate with Purpose webinar highlights that most firms are merely interested in AI because they risk falling behind competitors. From a critical-theoretical perspective—echoing thinkers like Marx, Jameson, and Piketty—AI thus reconfigures not only how architects work but may exacerbate existing inequalities in architecture and beyond.
The RIBA article’s emphasis on critical thinking aligns with this view. By foregrounding reflection and process, educators can resist a purely instrumental approach to AI that reproduces capitalist efficiency logics. Similarly, Deltek’s call for ethical AI governance suggests that technology should augment, not replace, the architect’s social and environmental responsibilities. Integrating AI into sustainable design, material optimization, or community engagement could redirect its power toward the collective good rather than corporate acceleration.
Conclusion
Artificial intelligence is currently reshaping architecture not through spectacular images, but through the quieter transformation of processes, data, and pedagogy. The cited sources urge firms and students to integrate AI strategically—grounded in data literacy, governance, and ethics—harnessing its productivity without sacrificing creativity or trust.
Is it possible that the upcoming advancements in AI for space planning and programmatic layout might uncover social configurations that architects have yet to explore? @MBA architects © 2025 Image created by AI assistant
Another step in the promise of AI (not yet fully captured by the industry reports), a proliferation of tools such as Architechtures, Finch3d, Maket, and Laiout that can parametrically generate floorplans and BIM models can perhaps fulfill the ambitions that Lionel March first set out in the 1970s. Although these tools are almost innumerable and largely cost-prohibitive, a rigorous study of their architectural production may reveal in their output programs, plans, and models social configurations that architects have yet to explore much like how AlphaGo made evident tactics never-before-seen in a centuries-old game.
In this light, AI becomes neither a threat nor a panacea, but a mirror reflecting architecture’s enduring task: to balance technological innovation with ethical imagination. The architect’s value in the age of intelligent systems has not substantially changed: it is and has been primarily in articulating questions, values, and spatial futures that machines alone cannot conceive nor understand.
Sources:
Alpaydin, E. (2016). Machine learning. MIT Press.
Architects’ Journal. (n.d.). How Architects Are Getting to Grips with AI: An Industry Survey Report.
Deltek. (n.d.). Innovate with Purpose: AI and Technology in the Built Environment [Webinar slides]. Panel with Tushaus, B. et al.
Keller, S. (2005). 10283EFE0F02 or the Seagram Building. In R. Hejduk and H. van Oudenallen (Eds.), The Art of Architecture/The Science of Architecture (pp. 298-306). Washington, DC: Association of Collegiate Schools of Architecture Press.
Morris, N. (2025, Oct. 30). What Can Architects Learn from an AI in Education Report? Royal Institute of British Architects. https://www.riba.org/work/insights-and-resources/professional-features/ai-professional- features/what-can-architects-learn-from-an-ai-in-education-report/
Picon, A. (2010). The Seduction of Innovative Geometries. In A. Picon, Digital Culture in Architecture (pp. 59-65). Basel: Birkhaüser.
Roose, K. (2022, Dec. 5). The Brilliance and Weirdness of ChatGPT. New York Times. https:// www.nytimes.com/2022/12/05/technology/chatgpt-ai-twitter.html
Tushaus, B., & Miller, M. (2025). How the AI Revolution Is Reshaping Architecture and A&E [YouTube video]. Deltek. https://www.youtube.com/watch?v=_H3twQM5F7Q
📷 Credit Images:
Comments