AI WON’T REPLACE ARCHITECTS – BUT IT COULD CHANGE THE WAY THEY THINK

Kanaka Raghavan, Senior Lecturer in Graphic Design, Middlesex University (MDX Dubai)

The way we design has always been a reflection of the tools available to us. As an undergraduate student about two decades ago, the tools we used were quite traditional: A sketchbook, a drafting board, piles of drawing sheets, tracing paper, freshly sharpened pencils, and a half-empty coffee mug. Design was physical, tactile, measured, and every revision cost you something. Technical drawings and models would take days to produce, and critique was received with extreme resistance due to the effort involved in making the slightest modifications. Gradually, digital drawing and 3D modelling software evolved to generating sections and elevations automatically, removing the long nights spent working on these manually. Having computers handle the repetitive, mechanical work has allowed one to pay attention to the most interesting part of the process: the designing itself.

In this mission to outsource dreary, tedious tasks to the machines, it seems like we may have gotten a bit carried away. With Artificial Intelligence (AI) already embedding itself into the mechanics of our daily life – the way we draft a message or edit a photo, it is inevitable for the technology to make its way into the creative process. Unlike conventional design software, which relies entirely on dimensions and constraints manually fed by the architect, AI has fundamentally changed the ideation process. Brainstorming concepts, generating multiple façade options, and exploring design possibilities has become far more fluid, like having a hardworking graduate assistant.

The architecture industry has seen rapid growth in AI adoption. Across early stages in areas like concept visualisation, specification writing, compliance checking, and product selection, as well as day-to-day practice management like report writing, bid creation, and project scheduling. According to a recent report, 59% of practices reported using AI on at least occasional projects, up from 41% the previous year, a sizeable increase.

Bigger studios have acted as early adopters. Zaha Hadid Architects developed bespoke AI software in collaboration with NVIDIA, exploring generative AI across façade patterning, structural optimisation, and urban-scale planning. For MVRDV, data is actively shaping building form from the very start. BIG’s Bjarke Ingels has spoken about how AI reduces the time between intuition and iteration, helping teams move faster through design options. Smaller design studios are finding their footing too. London-based Fu recently launched what has been described as the world’s first fully AI-driven architectural project, a residential scheme at Slovenia’s Lake Bled, where AI helped accelerate iteration and uncover spatial relationships that traditional processes might have missed.

Yet the technology, for all its speed and spectacle, requires scrutiny. Where AI has made its presence felt is primarily in the everyday operational side of practice – drafting emails, managing budgets, transcribing client meetings. Only 13% of practices are currently using AI for actual design and planning tasks. As Zaha Hadid Director, Nils Fischer, puts it, general purpose AI has a “pseudo-understanding of construction,” particularly bad at grasping how building elements actually meet and connect. While it is a capable assistant, AI is still a few upgrades away from playing a meaningful role in the construction process.

So, can the architect be confident about their future in the industry?

Designing buildings is like solving an intricate puzzle, and the architect has always occupied a unique position. Part researcher, part craftsman, part ethnographer, part philosopher. They aren’t just resolving a design; they are learning to understand people. To become a good architect, one must develop an empathetic read of the problem before proposing a solution. The creative process demands that we sit with complexity, allowing the concept to find its philosophical grounding before it becomes form. It is slow, sometimes frustrating, and entirely necessary.

That is where AI can pose a conundrum. When a tool can generate fifty design options overnight, how does the architect still develop the conviction to defend one? And how do they avoid the trap of sameness that AI has been accused of producing, where similar prompts fed into similar models begin to yield unsurprisingly similar results? Research in cognitive science suggests that exposure to others’ ideas early in the creative process can inhibit our thinking.  More unsettling still is the effect of passive AI acceptance. In a post-ChatGPT world, taking the first generated response and running with it has become second nature for many. Researchers are finding, however, that this habit may be rewiring how we think, not just individually, but collectively. And when individual creativity suffers, the diversity of our collective creativity inevitably suffers along with it. This is particularly worth considering for the next generation of architects, who risk treating AI as a silver bullet rather than simply another tool in their arsenal. While it makes connections we sometimes can’t, AI still remains a machine, not a problem solver, not an architect. There is still a vast distance between a striking AI-generated image and a building that can actually be built, specified, and guaranteed. The real skill lies in knowing what to do with that output, pushing it further, making it respond coherently against a brief, a site, a context.

An architect’s empathy is not merely a soft skill, it is the diagnostic tool that determines whether a building serves its people or merely stands. No model trained on images of buildings understands why a space needs to feel welcoming, or how a layout shapes the way people move through their lives. Those judgements cannot be prompted into existence. An architect’s value has never been in the mechanics of drawing, it’s in the thinking, judgment, and empathy behind it. AI can accelerate the process, but if architects let it do the thinking too, they risk losing the very thing that makes them irreplaceable.