Keir Regan-Alexander introduces a series of articles exploring four trends in AI that are emerging faster than expected

First Movers Title_Keir Regan-Alexander

Source: Keir Regan-Alexander

“Nostalgic Futurism 1/6”, created using Midjourney by Keir Regan-Alexander

At the end of 2022, OpenAI realised the most profound example of “product-market-fit” in tech history; when Chat GPT gained 100m users in just four weeks. In the short period since then, it has become clear from the accelerated release schedules at Google, Meta, Adobe etc that all the tech giants had been sitting on similar products for some time, but stuck within the “innovator’s dilemma” and not wanting to disrupt their own lucrative business models - they hesitated to launch them to the masses.

Sam Altman’s OpenAI were all too happy to open Pandora’s box to the public and now it feels like all the R&D that had been stored up for some future purpose, is now being dumped onto the consumer market all at once.

Like a growing number of architects I have been unable to look away - and we are now seeing a wave of business models coalesce around a particular set of ideas that are directly applicable to architecture, engineering and construction (AEC). The whole space is moving at a breakneck speed, with new versions and unpolished features being launched in beta just to get them out quick - it’s almost impossible to keep up.

To reduce some of the noise (and there is a lot of it), I decided it would be helpful to start tracking the most striking first-movers against RIBA’s ‘Plan of Work’. The plan of work, or “phases of design” (the American equivalent) is a predetermined roadmap for any construction project, tracking a project’s life from the moment of inception into the post-occupation of the building.

Each phase must be undertaken by the design team who fulfil specific project outcomes at each stage and this allows teams to organise their efforts around predictable workflows. To put it more simply, it means we know what to focus on and when. For a product to be relevant to AEC it has to know which part(s) of the project lifespan it is focusing on. In the chart below I’ve outlined a selection of 20 products that I’ve looked into and that seem to be offering something striking and each with different implications for our field.

Generative AI in Architecture_July 2023

Source: Keir Regan-Alexander

July 2023: A chart showing a range of 20 leading new products leveraging Generative AI innovations and targeted at AEC use-cases.

Perhaps unsurprisingly, startups appear very focused on the initial project stages where one would expect “Generative Design and AI” innovations to be most applicable. There are however, a few projects now focusing on the middle and later stages too, with the aim of addressing the more laborious and mechanical work architects undertake during drawing production and documentation.

Gradually, ideas are emerging about connecting directly to the supply chain for lower risk procurement and early costing – this perhaps points to the direction things are going. Over time I believe that each phase of a project’s life will be transformed in some way by this new breed of software application.

While most names on the list are purpose built AEC software tools, the adoption and use of more multipurpose tools like Chat GPT, Midjourney and Adobe Firefly is only set to spread and these are likely to become ubiquitous tools in the everyday work of architects, essentially serving as productivity assistants and ‘agents’ that work tirelessly under the instruction of the designer.

If you start each work task with the question “how could ChatGPT speed this up?”, you will be surprised at how many jobs it can help you with if briefed and directed in the correct way.

Over the next few weeks, I will be sharing a series of articles focusing on four key trends that we can observe emerging from the Generative Design & AI toolbox, and expanding upon which tangible problems they are aiming to solve. These will include:

1. “Auto Optioneering” - Where a model is programmed to test hundreds of massing and layouts arrangements against site constraints instantly.

2. “Hybrid Collage” - Where we can produce render quality visual experimentations by combining disparate image and text material.

3. “Leveraged Drafting” - Where we draw with whole building prototypes rather than line by line, using simple parametric controls.

4. “Pre-procurement” - Where we will be able to test procurement much earlier to review a design for offsite manufacture, early pricing and ease of construction.

It is often said that about 90% of all software start ups fail in their first year, but the 10% that endure long term from this current wave are likely to bring lasting change to how we work and will be followed by a more of less endless stream of VC backed projects that are now able to go after a market share that was previously well out of reach. Many architects will be aghast thinking about the idea of software solutions attempting to do any of the core work of architects and so we should consider these companies as sitting in two camps; those that aim to augment architecture practice and those that aim to replace a substantial part of it.

I am in favour of the former; of tools that work in partnership with the good judgement of an architects and that can improve the work of practices and reduce the amount of waste and low-value mechanical task work currently being done by architects.

But I am also of the view that the technological innovation we are beginning to see represents the start of a more profound change in context. It is against this backdrop that I believe architects need to react quickly and adapt.

>> Also read: Generative Design & AI Trends: ‘Auto Optioneering’

>> Also read: Generative Design & AI Trends: ‘Hybrid Collage’

>> Also read: Generative Design & AI Trends: ‘Leveraged Drafting’