There is a great deal of noise right now about artificial intelligence and what it means for the practice of architecture. Depending on who you ask, AI is either about to replace the architect entirely or it is a sophisticated autocomplete that cannot draw a decent stair. The reality, as is usually the case, sits somewhere more considered than either position.
This is not an attempt to offer a prediction, but a practitioner's read on where the profession actually stands, what AI is genuinely useful for, and why the creative act at the centre of architecture is not something that can be delegated to a machine.
A lot of noise, little action
For most architectural practices right now, AI is more discussed and experimented with than deployed in full, and while real tools are being integrated into actualized workflows, the profession as a whole is not in the middle of a transformation. The honest situation is that AI is beginning to appear in its strongest form at the edges of practice and will likely work its way inward, though the pace at which that happens is being considerably overstated, or so we see it as a means of our mode of practice today.
What AI is not is legitimately creative. There is no creative genesis in the AI tools and models available today; what they produce is, at its best, a well-presented synthesis of existing work assembled with little overall intent. That is not intended as a rebuke of the technology so much as an accurate description of what it tangibly does. While seemingly creative, AI is more factually just very productive. Creative and productive are not the same thing. In architecture, intent is everything. The difference between a building that pragmatically works and a building that affects a person's well-being, that provides an artistic lens, or one that provides space as poetry is always traceable to the moves made by human hands in service of a specific person, program, site, or idea.
Where AI can help
For the field of architecture, the honest answer is largely administrative. Repetitive workflow, schematic analysis, cost estimating, code navigation, documentation tasks, accounting: these are the areas where AI has genuine near-term value for an architectural practice, and for smaller studios especially. The administrative burden of day to day business can compete directly with the time available for creative work, and any tool that reduces that friction is worth taking seriously.
An analogy that holds up well is the arrival of 3D rendering and modelling tools nearly a generation ago; those tools did not change what architects could design so much as they changed how ideas could be effectively communicated and tested. Stakeholder presentations became more legible, schematic proposals that once required a trained imagination to pick up could suddenly be understood by anyone, and the idea was proposed as a believable iteration without the need to spend added time discussing detail that could be quickly modelled. The design idea itself still had to come from somewhere, and that somewhere was (and remains) the creative human mind rooted in an empathetic human perspective.
AI is likely to have a similar effect, expanding the range of what can be communicated and analyzed, accelerating certain administrative and technical tasks, and giving smaller studios more competitive capacity on larger or more complex work. What it will not do is generate the conceptual framework idea that provides a building its artistic basis, clarity, or refinement.
Architecture is not about creating imagery. It is a psychological pursuit, seeking a better approach to the emotional health of how humans inhabit space.
The question architects keep getting asked poorly
A common version of the AI-in-architecture conversation goes like this: if AI can generate convincing visualizations and schematic layouts, what separates an architect's work from what a well-prompted tool can produce? It is a question worth reframing, because it misunderstands what true architects actually do in the pursuit of a conceptually derived and reasoned building.
Architecture is not primarily a visualization practice. Architects work in the unknown, in the liminal space pinched between a physical reality and an idea that does not yet exist. The work is iterative, collaborative, and deeply specific to a person, site, city, or brief. It involves working through problems that have no precedent, negotiating between what a client wants, what a site will allow, what a budget permits, and what the city requires. None of that is a rendering problem, nor is it repetitive or precedential.
The more important question is not whether AI can produce an image that looks like architecture, because it largely can, but whether AI can produce a building in minute detail that makes a user's daily life measurably better. That requires understanding how humans relate to space, light, scale, texture, sound, and material at a level that goes well beyond pattern recognition, and it requires empathy that AI does not have and, in any meaningful sense, never will.
On democratization, and why that raises the stakes
One of the effects of AI lowering the barrier to visualization tools is that it may actually increase the importance of the things AI cannot do. When we get to a point where nearly anyone can generate a photorealistic image of a building, the value of an architect shifts further toward what sits beneath the image: the depth of thinking, the quality of the spatial sequence, the honesty of the material choices, the relationship between the building and the city, and the joy of place that building creates. The conceptual human input becomes more important as the surface becomes easier to imitate.
There is already a tendency for people to conflate architectural design with the production of attractive images, and AI will amplify that confusion. Part of the profession's response should be to articulate more clearly what the practice of architecture actually is and what it provides beyond instruments of service, drawings and renderings, and to make that case to clients and cities with more conviction than it has managed so far.
Clients arriving with AI-generated ideas
It is already becoming common for clients to arrive at a meeting having spent time with image generation tools or LLMs, and the result is more often a complication than a head start. AI presents its output as a solution, but it is more often an obfuscation of possibility, potential, and practice. Without flagging key concepts like structural implications, zoning effects, atmospheric effect, or the simple errors and oversights it invariably makes, it merely produces a convincing snapshot, and problematically, convincing snapshots can create attachment.
The work of the early design conversation has always involved understanding what a client needs as a function of their daily life and lifestyle, which is often quite different from what they think they want, or what an AI model will provide as a means of translation. AI-generated imagery or context adds another layer to that conversation: not just understanding the brief, but also gently unwinding the assumptions that a persuasive but ultimately shallow image has already installed in the minds of clients.
The role of AI in a small practice
Let's be clear: our studio leverages the strengths of AI. For a studio like Anonymous, AI is most useful as a support tool for the things that are, by nature, un-architectural: scheduling, documentation, analysis, repetitive administrative tasks, and some of the business overhead pieces that competes with time dedicated to creative generation. It works very well at baseline (but not in-depth) textual analysis, organization of data, and other administrative tasks. It is not a design tool in any meaningful sense, and it is not being treated as one here.
The bespoke, site-specific and client specific nature of our work is resistant to what AI currently offers on the design side. Every project begins with a client, a site, and a set of constraints that have never existed in quite that combination before, and the response to that combination must be the result of creative thought and design processes. Quantification, costing, and code navigation can be assisted by a well-targeted tool; the invention cannot.
The concern is not replacement. It is erosion — the gradual, barely perceptible thinning of architectural thought as shortcuts become normal.
What the profession needs to protect
The more considered fear about AI in architecture is not that it will take over the creative act, but that it will slowly degrade the depth of thinking across the profession as shortcuts become normalized and go unexamined. Buildings are too consequential, to our cities, our climate, and our health and wellbeing to be designed with such diminished care and attention.
The hope is that AI matures into narrow focused tools that specialize in specific tasks without colonizing the parts of practice that depend on full human engagement. Used that way, AI is genuinely valuable; but when used as a substitute for thinking, it produces exactly the kind of banal architecture the world already has too much of.
If you are a client thinking about a home, a renovation, or a building, it is worth asking the architects you meet where they stand on this. The answer will tell you something meaningful about how they approach design. What you are investing in when you hire an architect is not a set of drawings; it is a considered response to who you are, how you live, and what a specific piece of ground can become in the right hands. That is something AI can gesture at but never really produce, and it is what a practice that takes architecture seriously is built to do. If that is what you are looking for, we would like to hear from you.