How can your practice harness the potential of AI?

Get on board or get left behind – that was the broad message of an AJ webinar, run in association with project management software company Deltek, on the game-changing impact and opportunities of artificial intelligence in architecture.

Four industry experts sought to demystify the way practices can harness the potential of AI at a time of rapid change: Martha Tsigkari, senior partner, head of applied research and development at Foster + Partners; Keir Regan-Alexander, principal of Arka Works; Pablo Zamorano, head of geometry and computational design at Heatherwick Studio; and Bret Tushaus, vice president, product management at Deltek.

It’s a brave new world that’s both exciting and daunting. In the last year alone, 72 new AI tools have emerged, said Tushaus. He set the scene with an explanation of different kinds of AI: machine learning, deep learning, generative AI and algorithm-oriented AI. He sees three key areas of opportunity for architectural practices: ‘project deliverables’, for example use of AI-based technology to deliver more intelligent buildings; the process of project execution and design (including leveraging generative design and automating repetitive tasks); and tools to aid practice operations, such as AI assistants and predictive analytics.

He sees ‘tremendous potential’ for architects to ‘leverage the technology to boost productivity and creativity’, saying: ‘You’ll find different opportunities in different areas.’ He added that ‘thoughtful change management’ would also be required for smooth delivery with the least negative impact – AI could potentially replace some jobs.

According to Foster + Partners’ Tsigkari, AI and machine learning have the potential for unprecedented change in the profession.

‘Like Alice in Wonderland, we can either choose to ignore it or jump down the rabbit hole, where there is a lot of uncertainty but [where] there also exists a lot of possibilities,’ she said.

In addition to generative design, she cited the use of surrogate machine-learning models to aid predictive analysis, and also knowledge dissemination. Fosters has already introduced an in-house AI search tool for its design guidelines which uses natural language, and is now extending this to search drawings. The practice is also building its own generative machine-learning portal.

Creative minds shouldn’t fear that AI will replace them, she said. Instead, it will augment creativity and problem-solving capacity if appropriated correctly. Core to this is harnessing practice data.

‘The role of the architect remains crucial because it brings to the table not only creativity and innovation but also aesthetics, emotion, collaboration and ethics coupled with responsibility,’ she said.

But she warned that much work needed to be done more widely on AI rules of engagement in relation to data contextualisation, regulations, IP, education, ethics, and dealing with unconscious bias.

Regan-Alexander, formerly a director at Morris +Company, recently set up Arka Works, a creative consultancy exploring the application of generative design and AI tools in the built environment. He shared a case study developed with Mae Architects to eloquently demonstrate just some of the uses for AI. Using a generative massing tool, this created a first-pass masterplan for 1,500 homes with suggested housing typologies, areas and block depths for the designer to refine. Other tools enabled scope for client and designer to experiment with different apartment variations and gave environmental analysis and embodied carbon predictions to assist the designer as they made key decisions at an early stage.

‘It’s a more fluid and direct route to validation,’ he said.

He cited many other applications of AI already used in practice, including bid writing, marketing, data processing, code writing for interoperability and copy editing. He anticipates many further future uses including smart intranet, fee estimation, legal review, building compliance, and live pricing, engagement, and carbon calculation.

Zamorano, of Heatherwick Studio, saw opportunities for AI in both design and knowledge sharing. He shared how his practice had been exploring how to use AI to get an outcome that was ‘à la Heatherwick’ rather than generic. Key to this is leveraging its own data and images to train models.

‘Tools like this, for us, are great provocations for starting design conversations and taking them forward,’ he said, adding that the practice was keen to engage with others in the industry on AI use. ‘We’re pretty much all learning.’

One of the downfalls of AI, he added, was that’s very easy to lose control of what you’re doing. ‘It’s so key to understand what you want to get out of whatever you’re doing with the technology,’ he said.

So how should practices get started on exploring how they can best use AI?

Alexander-Regan advocated a practice working group that goes beyond just the tech people to develop a strategic plan.

‘It’s an organisation-wide issue.’ he said. ‘You need to think about your whole processes and you need leadership involved … It will touch all parts of your organisation, so you need a holistic view.’

Tushaus agreed. ‘Start with a taskforce, a group that’s very diverse in terms of experience and education, with dedicated time to determine what a firm can leverage from an AI perspective,’ he said.

Tsigkari advised practices to consider how they manage their data, and understand the IP implications – ‘a huge risk’ – of using certain AI tools, and the different ways that the models they use have been trained.

‘Know what data you have and organise that data properly. Have an AI legal framework in the office. Ensure that you understand what you’re using and for what purpose,’ she said.

Zamorano advised practices to fully engage with AI. ‘Go for these tools – they won’t go away – and learn about them and what they do, how they work, and think then what you need to use them for. And totally embrace them.’

In association with