The Ethics of AI in Architecture

An LA London team member using Archicad during the design process.

As architects, our role is not just to design spaces that look good and function well. We are also dedicated to safe and sustainable practices, and to the wellbeing of our staff, clients and partners. 

A survey by Architects’ Journal revealed that around two-thirds of architects currently use some form of AI, the most popular tools being Chat GPT and Midjourney. These tools can help with concept and administrative tasks like creating visualisations, generating images, and communicating with clients, consultants and contractors.

As it becomes more sophisticated, however, AI may be used for more complex analytical and project management processes, which raises concerns around safety, sustainability and responsibility.

As part of our journal series exploring the role and evolution of AI in architecture, we’ve taken a look at some of these ethical considerations, outlining how we intend to use new technology in safe and responsible ways.

Our collaborators, Lightfield's AI-generated imagining of Gustave Eiffel's desk

Safety

Like the human intelligence on which it is trained, artificial intelligence is imperfect and sometimes prone to errors. In a worst-case scenario, this could undermine the safety of a building and its occupants. 

Rigorous project monitoring is fundamental to our approach at LA London – so much so that we are frequently asked to perform the role of independent CMT (Client/Compliance Monitoring Team) architect for other firms’ projects.

We also established our own dedicated measured surveying arm, LA Surveys, in order to address the inaccuracies we had been seeing on surveys prepared by others, which could cause problems and delays at the later stages of a project.

Precise reporting, comprehensive record-keeping, and continuous risk mitigation are key to the successful and timely completion of our projects. We see AI being used in the near future to support these processes, closely supervised by an experienced team member.

An AI-generated visualisation for a recent LA London  refurbishment project

AI-enhanced tools allow us to measure and model spaces more accurately, ultimately resulting in a safer product. Our go-to design software, Archicad, has already started implementing AI capabilities that put more resources in the hands of the experts.

AI can also be used to simulate a building’s behaviour during a fire or extreme weather event, allowing architects to spot vulnerabilities and address them in order to keep occupants safe.

Copyright

Architects have long sought inspiration from existing buildings when designing new ones. The Blackpool Tower, for example, was openly modelled on the Eiffel Tower, built five years before. Irish architect James Hoban’s design for the White House was based on Leinster House in Dublin.

Today, practice archiving is a method by which local AI models are trained on the architectural style of a particular architectural practice. This raises concerns about copyright and IP: authors and artists have already filed lawsuits against companies including OpenAI and Stable Diffusion for using copyrighted content to train their models. 

Earlier in the year, the House of Lords Communications and Digital Committee called on the government to protect copyrighted materials used in AI tools, and we would certainly welcome action on this front.

Bias

Relying on AI introduces concerns around algorithmic bias – whereby models are trained on data that lacks diversity. This could have implications for a vast range of regulatory, cultural and environmental factors. It could also negatively impact the end-user by not addressing accessibility and safety needs.

It’s therefore our responsibility as architects to ensure our projects continue to meet and exceed the unique requirements of our clients and to comply fully with local regulations.

Trust

Trust is essential to successful relationships with clients and consultants, and between team members.

One of the concerns about integrating artificial intelligence into the design and construction processes is that it undermines that trust, particularly when it is used covertly or to cut corners.

We have always been proudly transparent about any new technology we use, whether it’s our early adoption of BIM or our expertise in cloud point surveys. Using the same or complementary systems as consultants and engineers has also enabled us to work more closely and collaboratively with them.

In addition, our ability to produce more lifelike and accurate visualisations helps our clients to see more clearly how a project will look, ensuring we are completely on the same page and supporting the culture of trust and transparency that is so important to us.

Employment

It’s natural that many of us are concerned about the impact of AI on our jobs. A report by Goldman Sachs predicted that around 10% of architects’ workload could be replaced by AI automation. In response, we might argue that we would happily relinquish that 10% and still be fully employed!

None of us can be sure what the job market will look like in five or ten years’ time, but it’s safe to assume it will look different: some existing functions may have become obsolete, while others will have been newly created.

“Our talented team is by far our greatest asset, and we’re delighted that it is growing at a time of such rapid technological innovation.”

All we can do is look to the past, where major technological advancements have generally not reduced the demand for humans but allowed us to become more productive and efficient.

At LA London, our talented team is by far our greatest asset, and we’re delighted that it is growing at a time of such rapid technological innovation.

Sustainability

AI-powered simulations may contribute to our sustainability efforts. Improving the accuracy of our models means we can use the most appropriate materials in the correct volumes, cutting waste and reducing carbon footprints. Energy simulations that use AI also allow us to refine heating and lighting systems, reducing energy consumption. 

Archicad’s embodied carbon calculator already allows us to see the embodied carbon built into our designs throughout the design process so that we can consider using more sustainable products, materials and systems if necessary.

Potential led by principle

The potential of AI in architecture is huge, but we believe its use should be guided primarily by our values and by our clients’ needs. 

We are cautiously excited to explore new tools, which we will integrate in fully transparent and carefully documented ways, and as long as they align with our commitment to safety, sustainability, and to our staff and clients.