AI For Architects Is Nothing New. And It's A Return to First Practices (Finally).
Tools like Veras are not another evolutionary hurdle in design visualization. AI represents the opportunity to return to first practices. A restoration of the original order.
Architects have traditionally begun with the Sketch: A napkin, a pen, a composition. The emotive Sketch was the prerequisite to the rational Design, not the other way around. But for the last thirty years (or so), CAD and BIM reversed the sequence, where you started by creating the model and then rendering as a byproduct.
What if sketching is an essential first part of the design process? A way of discovering, with the lowest technical barrier to entry: draw a line with one end of the pencil and delete with the other. Conversely, CAD, 3D Computer Modeling, and BIM became the realm of the technical specialist. How deeply unfortunate our most experienced designers were frequently alienated mid-career with choosing between often irreconcilable skills: pen and trace vs. keyboard and mouse. Such were the artificial constraints of awkwardly evolving technology.
AI restores the original order of sketching > design. AI is not about going fast. It's about going slow and being more deliberate, empathetic, and gestural. Only when the idea is ready do you commit to the computer. Or in the case of AI, upload and prompt. The Architect's struggle (fear of failure) remains, yet relocates. To be more inspired, rigorous and accountable given constraints of time and budget. Why this form, this space, this orientation, this material, this light. AI is readily available to test and confirm if a concept is a worthy vector (or shite).
Overcoming Objections.
1. AI is new and difficult to comprehend. AI has always existed in the original, carbon-based form of the Architectural Intern (AI). The lead designer meets with the design team. They discuss constraints, sketch, deliberate, debate, ideate, describe. Then the architectural interns and design team departs to bring life to vision. Now the AI is silicon based. Not replacing, but augmenting. An Intern for every Intern.
2. AI will eliminate jobs. AI will exponentially expand visual communication. Upstream to the familiar pen and paper of the analog designer and downstream to the engineers and contractors. Both use cases are historically too time-constrained to become visualization specialists.
3. AI is another specialist role. From chauffeurs to cell phones, every technical evolution expands from the priestly specialist to the lay generalist. Each role remains valid depending on circumstances. Consider the expanding role of architectural visualization of the last few decades:
20 Years Ago: V-Ray: For the deep technical specialist who learned how to model and render. Geometry, lighting, materials, set dressing, entourage, post production, and more. The final product requires decades of experience.
10 Year Ago: Enscape: For someone who knows how to model, but doesn't know how to render. Geometry, lighting, and materials are a visual byproduct of the concurrent design process.
Present. Veras / AI: For someone who doesn't know how to model or render. Yet approachable enough for anyone who can sketch, compose, ideate and describe with judgement.
4. AI produces derivative work. Humans have always remixed centuries of precedence; AI does the same. How often you've searched the web for images and inspiration, dog-eared books and magazines, and photographed architectural details for later reference. Lazy prompts create mediocrity. Prompts are poetry. Elegantly describe climate, cultural references, intention, and combine with your own hand sketches. AI has the potential to generate well beyond an ill-conceived transaction. Form follows function; Pixel follows prose.
5. AI hallucinates. As does your carbon-based intern. A misunderstood corridor becomes a thoughtful courtyard. Sometimes you discard; but occasionally you explore the unintended accident and build something better. Hallucinations aren't bugs—they're opportunities.
6. AI short-circuits education, judgement and rigor. Careers happen when you're willing to be accountable for your decisions. Embrace AI in architectural education and business. However, demand students and interns follow through and remain accountable from emotional image to empirical effort. There is discovery and wisdom in rigorous rationality. Embrace the journey from anything to something. AI may allow you to suspend judgement, but eventually gravity gets the last word. Own it.
Present Limitations.
AI for Architecture Design is too transactional. Prompts are one-way demands rather than open discussions. AI for Architecture should be iterative and voice driven. I yearn for AI Agent mentors (alive, unavailable, or long gone) designed to deliberate, ideate and debate. Why limit design critiques with an available professor? What would Steve Jobs say? Marcus Aurelius? Tadao Ando? When a designer settles on mediocrity, perhaps the AI should respond with, "I'm sorry Dave, I'm afraid I can't do that".
This essay exemplifies the previous paragraph. Rather than starting from a blank screen, I tried a more dynamic approach. Pacing back and forth, I spoke aloud, trying to capture big ideas in a stream of consciousness. Grok, my preferred AI assistant, served as a digital counterpart: listening attentively, challenging assumptions, synthesizing scattered ideas, rearranging segments and pushing back with counterpoints that required more deliberation and refinement.
Grok and I engaged in an iterative dialogue, similar to a conversation with a trusted colleague. For instance, we explored the initial premise that AI represents a return to first practices of Sketching and Rendering in order to Design. We also considered potential objections—such as job displacement, the risk of soulless designs, hallucinations as creative sparks or pitfalls, and the perception of AI as “architectural fast food" - hastily prepared and consumed - rather than crafted and appreciated.
Through voice driven back-and-forth exchanges, we addressed concerns. We discussed countering fears of job loss by noting how tools like V-Ray, Enscape, and Veras increased access without eradicating judgment; why reframing hallucinations as opportunities are akin to an intern’s misunderstanding that yields unexpected opportunities; and ultimately how technology expands beyond the dedicated specialist. Complicated features eventually give way to elegant ease of use. Back to first practices of pen and paper, then render and resolve.
One More Thing™.
CAD, then BIM, and now AI!? Here we go again!? AI for Architecture isn't another technical specialist hurdle to overcome or risk getting trampled in a hype-cycle of enthusiasm over experience. While present iteration lacks subtlety, AI is the Architects opportunity to return to first practices of sketching in order to design. Of course, every technical hype-cycle comes with a cautionary tale; wings are wonderful, but take care not to fly too close to the sun; iPhones connect us, but how easily we are distracted by the least common denominator, envy and doom scrolling. Architects should embrace the opportunity to integrate AI as a digital version of the Architectural Intern (with similar pitfalls, surprises and successes). And for once, the interns will have interns.
Chaos Talks AI. As Chaos Community Ambassador, I'm privileged to host the "Chaos Talks AI" monthly podcast with thoughtful industry peers. We're not just "drinking the kool-aid". In my experience, we're cutting through the AI hype and discussing what works, what doesn't, and why AI is most often useful for people who know how to do something well with analog tools. The nice thing about the podcast is that we have live community chat for questions, concerns and downright objections. Please join the discussion and register over here: https://www.chaos.com/chaos-talks-ai
Chaos Talks AI Monthly Podcast
Why Does It Matter? Before Architecture, I studied at seminary. It wasn't long before realizing one generation's heresy often becomes the next generation's dogma. Both are unfortunate extremes. Similarly, those arguing you must never use AI, and those insisting you must always using AI are posing reductionist extremes, which lack nuance. My sense is this "new" AI technology is difficult to comprehend until you realize it's not new. There's always been AI: It's called "staff". Think of silicon-based AI as the digital version of long familiar carbon-based Architectural Intern. Both are capable of the surprising, the unexpected, and even the undesired. Then what? It's still up to us to provide judgement.
(Footnote for another day: If you think seminary can be dogmatic, try studying Architecture.)
More from the author → Phil Read on LinkedIn
