Working Smarter? Initial Reflections on AI in Architecture

Human – AI – human. That’s our workflow. Artificial intelligence has entered nearly every industry, and architecture is no exception. Over the past few months, our team has explored what AI might mean for us, testing tools like ChatGPT, Midjourney, Nano Banana, and Spacio AI. The experience has been equally fascinating and humbling.

As with all new technologies, there’s a temptation to overstate or underestimate its potential. But so far, we’ve found that AI, at least in its current form, is neither a miracle worker nor a threat to our profession. It’s simply a tool that, when used thoughtfully, can help us work smarter, but not one that comes close to replacing the depth and nuance of an architect’s eye and mind.

In August, we attended the AI Empowered Conference, which offered a wide-ranging introduction to this rapidly expanding and still-surprisingly analogue field. The sheer pace of change is staggering. Every few weeks, a new platform emerges, claiming to make our work faster, easier, or more efficient, or simply to increase the risk of distraction. The more we experiment, the clearer it becomes that the power of AI lies not in its independence, but in how it amplifies the skill of the person using it.

AI is an amplifier. In the hands of someone experienced, someone who already understands proportion, context, form, function and budget, it can help speed up the mundane and spark fresh ideas. It produces a lot of output, and a discerning mind can sift through it, picking out useful fragments while discarding the noise.

However, AI can easily become a distraction for those still developing in the field. Without a trained sense of what’s viable, practical, or meaningful, it’s easy to follow a trail of digital breadcrumbs to a naive end. The technology may be capable of producing endless iterations, but it cannot yet discern what’s good, appropriate, or buildable. That still requires experience, intuition, and a sense of place.

The same applies to the written side of our work. The speed at which AI can generate structured, logical text is remarkable for reports, assessments, and presentations. But as we’ve learned, it often produces arguments that sound intelligent but may lack depth or contain subtle contradictions. An experienced professional can catch these inconsistencies and use AI as a productivity aid. Someone less seasoned, however, might take the output at face value and end up presenting something that doesn’t hold water.

Architecture is both technical and profoundly human. It’s shaped by context: physical, social, and emotional. Machines can simulate structure and render beautiful images, but they cannot walk down a street and imagine the experience of arriving at a doorway, and the visceral anticipation of the person knocking on the door. They cannot sense how the afternoon light softens through a north-facing window or feel when a room’s proportions are slightly off. Context and emotion are the invisible layers that define our work, and these remain beyond the reach of any algorithm.

At the same time, it would be naïve to ignore AI. Like every other industry, ours must learn to work with the tools at its disposal. The challenge is to do so in a way that enhances rather than replaces our humanity. We can utilise these tools to eliminate friction where possible and allocate more time to designing spaces that move, shelter, and inspire.

The gift of AI isn’t to outsource creativity but to give ourselves more time and headspace for it. Like any new material or technology we’ve ever adopted, it will take time to understand where it truly adds value. For now, we remain curious, critical, and open to approaching AI with the same balance of adventure and thoughtfulness that guides all our work.