How A.I. Will Change How We Build in 2025
Design in ‘25: The Near Future of Architecture & A.I., Pt. 3
In This Post:
Unpacking 2024 . . .
In 2025 . . .
GCs Have a Very Big Year
Sustainability, Finance Drive Adoption of New Materials
We Normalize A.I. on the Jobsite, Trailer
America Gets Serious about its Housing Problem
Unpacking 2024 . . .
In 2024, I had a couple of predictions about how A.I. will change the way we build, with a general theme: in the short term, the infusion of A.I. will aggravate the longstanding professional conflicts between AEC professionals, but thereafter, A.I. will enable unprecedented levels of coordination and likely a blurring of professional boundaries.
Underlying these projections was my belief that generally, contractors will adopt A.I. faster than architects, only because the benefits for contractors are more readily monetizable.
We haven’t seen any radical, sci-fi job sites yet, but that’s likely because design/construction cycles just take a while. My insiders tell me that many future-forward technologies are already being eyeballed, and the chief constraint at this point is project cycling and liability. GCs have to wait until a new project comes online, and that new project must be the right fit.
In 2025 . . .
In 2025, I’m looking at a few other trends:
GCs Have a Very Big Year
I think that 2025 is going to be a good year for contractors, regardless of whether they adopt A.I. or not. New Housing demand is going to rise (see below), and varying rates of A.I. adoption will play into the contractor’s favor.
Regardless of what A.I. adoption contractors do, there’s certain parts of the construction process that can’t be sped up – like the curing of concrete. If design processes speed up faster than construction, you get a backlog. If it takes you 30 minutes to make a pie, and 30 minutes to bake one, you have 1 pie coming out of the oven every 30 minutes, or 2 pies per hour. If process improvements allow you to make a pie every 15 minutes, you still only have 2 pies per hour, because it still takes 30 minutes to bake each pie – that’s the choke point. What you get is a backlog of unbaked pies.
A.I. is going to speed up real estate development, predesign, design, etc. a lot faster than its going to speed up construction. As that backlog grows, demand will increase for contractors, driving up wages.
If Donald Trump is successful in his stated plans for mass-deportations, that could make the situation worse, given that 1/3rd of the construction sector’s labor force is foreign born. This would mean higher prices for home buyers, but higher wages for construction workers.
Sustainability, Finance Drive Adoption of New Materials
I've written previously on the enormous amount of new material discovery currently being performed by A.I. Discovering new materials is by no means simple, but it is procedural, and is therefore fertile ground for A.I. Google’s DeepMind’s Project GNoME exploded conventional methods of material discovery by discovering hundreds of thousands of new materials at one go.
However, the materials that we build with are institutionally entrenched through vast networks of supply chains, production systems, and cultural know how. So just because something’s been invented, doesn’t mean we’ll see it on a job site.
My guess is that architecture will evolve as it always has . . . There will be a few practices out in front, playing with, and experimenting on, the new materials, and it will take the rest of us some time to get accustomed to their use, and their aesthetic. It will take time, as well, to adjust supply chains. If you consider that reinforced concrete was invented in 1848 and didn't come into widespread use until after the First World War, you know that just because a material works well doesn't mean that it's immediately put to use. Now that material discovery seems sewn up, we’ll prioritize adopting materials that facilitate sustainability and financial goals, rather than aesthetic ones. Practices that find low-impact ways to experiment early will be at an advantage when the materials market starts to evolve.
We Normalize A.I. on the Jobsite, Trailer
I previously referenced a video by Ethan Mollick which alarmed me because it showed how a non-professional could use A.I. to start to do professional-level things on a job site. For reasons of liability and licensure, I don’t think that’s an immediate threat, and in the meantime, AEC Professionals need to be equally creative about how all this new technology can be used.
For instance, why not replace the construction trailer with drones? No, really, I'm serious. What about, rather than having a construction trailer staffed by full time supers tweaked out on bad coffee, the site just had a standing fleet of drones that would hover over, in, and through the site on a continual basis? DARPA already does it. Using coordinated land-based and aerial drone swarms, one could record everything and provide project executives with all the eyes on the project that they needed. Machine learning algorithms could evaluate the site and predict problems before they happen. Sounds fantastical? Only a few months ago, a team of doctors at Mt. Sinai used video feeds and machine learning to predict neurological disorders in newborns. The A.I. outperformed human doctors in correctly identifying neurological disorders merely by watching the babies sleep.
If there were a serious problem, a human could always be dispatched to the site, of course. But would turning the jobsite into a panopticon obviate the need for the traditional construction trailer? I'm not sure - I'm not a GC. But I hope some GC is thinking about it. And I would look to see GCs start exploring this by proposing lower personnel numbers and higher technology costs.
America Gets Serious about its Housing Problem
My intuition is that recent trends in housing undersupply and unaffordability will start to reverse. Maybe that’s just wishful thinking from someone who’s trying to buy a house. But my reasoning is that housing has simply become an untenable political problem, evidenced by the fact that both 2024 Presidential candidates offered specific and dramatic proposals to address the housing crisis.
The U.S. housing problem is complicated, but one feature is the so-called ‘lock-in’ effect: homeowners who locked in ultra-low mortgages in the years after the Great Recession are unlikely to want to move and take on a new mortgage at (relatively) exorbitant rates. According to the CFPB, nearly 60% of the 50.8 million active mortgages have interest rates below 4%. For those homeowners, only a dramatic drop in mortgage rates would make selling a net win.
It seems unlikely, therefore, that a market-based solution is at hand, which points to either a technology-based solution or a policy-based solution. Realistically, it’s both.
Technology-based solutions, like 3D printing and modular construction, offer a lot of promise and have made great strides in recent years. However, any technology-based solution is useless unless the policy prescriptions are in place to support their adoption. Zoning laws and building codes were written around the technology of their day, and often need to be rewritten before a new tech can take off. The modular construction industry has been making great strides on this through dedicated advocacy. To date, Virginia, Utah, Montana and Rhode Island have all recently adopted ICC/MBI 1200 and 1205 standards for offsite construction. And FEMA incorporated modular housing strategies as part of its plan to rebuild after the Maui wildfires.
If you enjoyed that, be sure to check out Part 4 of this series: 'How A.I. Will Change Architectural Clients in 2025' and subscribe below for all future updates.