ConTech Conversations presents monthly discussions with construction pros who are using technology to improve the way their companies do business. Click here for past conversations.
Aaron Anderson is staying grounded on artificial intelligence.
The director of innovation for Concord, California-based builder Swinerton, Anderson talked with Construction Dive about the technology and more at the ENR FutureTech conference in San Diego last month.
In his role at Swinerton, he’s developed technology for Swinerton’s Perq parking structures, which provide clients with a pre-designed and pre-engineered garage, according to the company’s website. He also contributed to the growth of the firm’s mass timber, concrete and drywall divisions.
Here, Anderson talks with Construction Dive about a few key ideas on generative AI’s current capabilities, while also looking to the future on what could come next for the technology.
Editor’s Note: This interview has been edited for brevity and clarity.
CONSTRUCTION DIVE: What did you hope to learn at FutureTech?
AARON ANDERSON: It's always important to understand how everyone else is using the tools that are out there. Being at a company like Swinerton, those tools and startups and more established technologies find us, and we hear about them.
There are three key areas that I'm looking at that will heavily affect our business.
One of them is AI tools, and how folks are finding something beyond basic large language model applications, how they're actually really using them for the construction process and in some of these special use cases in construction.
Then there's robotics, and how people are finding ways to work those into their business models.
The other side of it is, generally, how people are using technology to shift the business model to provide new services. It’s about new ways of working with clients, but also new ways of taking this core competency around construction and turning it into other businesses.
Are you working on any projects right now in those three focus areas?
We're working on a number of things with AI. We're a Microsoft shop, so we have access to a lot of that stuff, the OpenAI tools that are hitting Microsoft already.
Truthfully, they've been doing a lot of interesting stuff for a long time when it comes to recommendations and design ideas and all that. But none of that really gives us an edge.
When we look for an edge, we're looking at things like project management processes, or looking to really dig into the ways that we would process things like submittals or RFIs, and the way we communicate about what's happening with design changes.
And, we’re looking at the way that we turn requests, and the work and expectations from our client into daily reminders, how to track and keep up with the expectations that we've set for projects.
What’s it like being a Microsoft shop?
We use the Teams tools, we use Outlook. As a Microsoft partner, we also have access to the Copilot stuff that they've changed. They branded that as Bing Chat and some of these tools like that.
Those Copilot tools can work in a bunch of different applications. They can work in their Power Automate platforms, they can work in their Excel platforms, they can work in their Word and PowerPoint platforms. But we've really struggled to find a way to make a major impact on that outside of our marketing and knowledge management.
It will make an impact on reviewing large, complicated documents. And then, managing that and turning it into actionable steps.
I think it's going to be in things like scope from construction documents, how to define all the work that's presented, and then be able to make sure that systematically you can move through that.
In the long run, I think there’s the opportunity to take all of these documents that are in a digital format, but not digital ideas. They're not uniquely sorted into rows and searchable and groupable and manageable, digitally. There’s opportunity in turning those into things that are manageable digitally, and making them into things that you can get to sort and filter and group and combine those specific requirements of a project or those specific areas of work.
What do you think about AI as a solution for the construction industry?
AI is great for things where there's a high level of uncertainty. What AI tools, right now, are really driving the construction folks to do, is to catch up on putting the systems in place to have answers to those finite solutions.
But sometimes they're big, and they're really complicated solutions. AI alone isn't going to solve construction problems. But, faster, more available compute might.
There's so many different systems at play that if you could observe them and train on them, that would be great. But we're constantly running into brand new things, brand new ideas, brand new ways of building, and those tools would have to be able to train on all of that historical data, and it's just not available.
We've already seen the fact that large language models are great at things like marketing, but are still not aware of physics.
I think the biggest thing that it presents an immediate opportunity for is the fact that at each stage of construction, we hand everything off in PDFs. It would be really helpful to have that stuff with a more proper digital connection with unique identifiers that connect those things from space to fit space. And I think right away, AI tools offer us an opportunity to do that.
