With the emergence of AI being one of the dominant themes of the 2020s to-date, architecture is one of many industries grappling with this technology, its capabilities and future implications.
On the one hand, we’ve found AI to be a powerful design tool that can be used to broaden a professional’s capabilities. And it will only get better with time (up to a point, at least — more on that later).
On the other hand, we have serious concerns about the widespread disruption AI may bring.
So is AI in architecture really a modern Industrial Revolution, or is this yet another tech-fuelled bubble?
Read on for our take.
How architects are using AI today
While the technology is still in its infancy, architects are already experimenting with AI in their workflows:
Workflow efficiency
AI is excellent for creating quick rendered images to understand various design or colour options. This frees up more time for architects to strategically plan and focus on high-value activities.
Rapid design and ideation
Whether it’s ChatGPT, MidJourney or an AI integration within specialist software, AI can generate variations in minutes. These can be used to initiate conversations and offer fresh perspectives, but a lot of checking and refinement is still required to get a design to a presentable, practical state.
Boosting productivity and scalability
Armed with their new AI ‘design partners’, architects can use AI to scale design iterations, visualisations and presentation material faster. But we advise caution when considering ‘faster’, as the primary goal of architecture is not to create something quickly, but a unique design with staying power that is on-brief and on-budget.
Increasing client engagement
Clients can use AI to visualise their ideas more clearly and experiment with different finishes. For example, we had a client send back our design for their new kitchen with different materials, all generated with AI. While this does shorten the feedback loop, it also adds complexity, because being able to generate a concept doesn’t give you the expertise necessary to analyse a design holistically and understand why all the pieces are as they are.
Myths about AI in architecture
Despite the hype, we’ve seen firsthand that there are plenty of misconceptions about what AI can and can’t do in architecture. Here are a few common ones.
Myth: ‘AI will replace architects’
Reality: In our informed view, it can’t.
Don’t get us wrong, AI is already reshaping parts of the architecture industry. But AI being used entirely in place of an experienced architect? That’s unlikely. Architecture is a multi-faceted discipline including complex stakeholder, budget and risk management — generating attractive images is just one aspect of what we do.
Myth: ‘AI delivers ready-made results’
Reality: You must check AI outputs very carefully.
You’ve probably heard the old disclaimer: actual results may vary. And it’s true of AI. Because while AI tools are great at producing slick visuals, the output may be impractical, and worse, fraught with IP concerns as AI tends to borrow heavily from existing content. Without a real brief and context around site conditions, budget and code compliance, AI’s outputs can be misleading and completely unbuildable. What if that shiny AI render with nice big windows and futuristic lines is not achievable within your site/budgetary constraints?
Myth: ‘AI stifles human creativity’
Reality: On the contrary, we’ve found that AI can actually complement and boost your innate human creativity.
When a tool lets you shave down the dull, time-consuming aspects of a task, you have more space to think and refine your approach — this is the root of all great creative output.
Myth: ‘With AI, designs are cheaper’
Reality: While the most obvious commercial benefit of AI is its potential for cost/resource savings, that’s not the full picture.
Because those efficiencies come with trade-offs. Adopting AI means new software subscriptions, re-training staff, and more time reviewing outputs for risks including liability, quality and copyright.
Myth: ‘Anyone can be an architect with AI’
Reality: This is a big one. And the short answer is, no, being able to prompt an AI doesn’t make you an architect.
AI is excellent for generating visions, but architecture is about creating visions and then making them into realities. That means site constraints, engineering challenges, compliance, navigating local planning authorities, and above all, ensuring the home you’ve designed is safe, fit-for-purpose and gives the client the feeling they want. Put another way: would you trust someone to be your town planner simply because they’re good with Google Maps?
AI architecture: benefits at a glance
Speed and efficiency
AI expedites early-stage workflows like rendering which enables architects to iterate faster and explore more options before locking in a direction. Speed alone is nothing without rigour, though.
Diversity of output
With the ability to generate many variations from a single brief, AI tools can expand an architect's creative output, help unlock new forms and challenge assumptions. But remember, often less is more. Being presented with a couple of well thought through concepts is usually preferable to a deluge of half-baked ideas.
Equity
Smaller firms can now use AI to gain in capabilities that were once the preserve of large, well-resourced studios. This includes image generation and simulations. In this way, AI lowers the barrier for innovation, but it takes more than flashy concepts to be a top architectural firm.
The ethics and risks of AI in architecture
Accountability and liability
AI doesn’t carry liability, architects do. If an AI-assisted design ends in failure, the responsibility lies with the architect. That makes human oversight essential, not optional, even when tools appear supremely confident or convincing.
Over-dependence on AI
If you become over-reliant on AI, your core design skills and critical thinking are at risk of eroding. We think it’s critical that like with any tool, AI should support and not replace your own decision-making.
Environmental cost
AI’s carbon footprint is rarely discussed, but it's real. There is a heavy computing burden and massive server infrastructure required to train and run large AI models. While this statement from OpenAI’s Sam Altman suggests the energy consumption of a single ChatGPT query roughly equates to operating an efficient lightbulb for a couple of minutes, keep in mind that ChatGPT fields billions of queries every day. Computer chips used in AI are also subject to 'swift obsolescence', as publicly acknowledged by some of AI's biggest companies, leading to significant e-waste. This means for firms serious about sustainable design, there is some tension between their ethical goals and their use of complex digital tools.
AI might not get as good as we assume it will
We’ve become used to developments in AI moving at lightspeed. From the moment ChatGPT burst onto the scene, it’s been regularly upgraded and countless competitors have joined the fray, all vying to out-genius each other. But there are signs that the rate of improvement is already tapering off, so it’s by no means a given that AI will become some omniscient force. If we’re already nearing the point of diminishing returns, it could be that AI in future really is just another tool in the kit, albeit a novel one, so we need to prepare for that outcome.
TGA perspective: ethically applied, carefully monitored integration of AI into the workflow
At Team Green, we've started testing AI in our workflow in a careful, contained way. We’re ensuring we use it thoughtfully and transparently, and never without proper human oversight. We’ve found good, simple uses for AI tools include generating material options for interior design, or cladding options for a house.
In our consulting practice, AI supports better data analysis, allowing us to identify patterns, risks and opportunities that inform project strategy. With better data, we help clients make more informed decisions, not just faster ones.
Conclusion
AI may be changing how we work, but it doesn’t change what architects do. View it as a powerful collaborator, but know that it lacks a professional architect’s emotional intelligence, critical thought or real-world expertise. AI may be a very impressive new type of pencil, but it’s not an architect.
