AI in AEC: Can we Revolutionize Better this Time?

This past November, I was delighted to join roughly a thousand fellow professional structural engineers in Las Vegas for the annual summit of the National Council of Structural Engineering Associations. I took the stage for the keynote panel alongside Rob Otani and KP Reddy to discuss the future of our profession in the era of AI. The talks that morning kicked off three days packed with sessions, interviews, and casual conversations, all demonstrating how exciting and alluring the topic is right now. So much so, that attendees, weary from wandering the endless interiors of Vegas casinos, were more than willing to tolerate standing-room-only conditions to get insight into corporate onramps for and legal conundrums presented by the technology.

Of the many people I spoke with, most were optimistic and fascinated. But there were also plenty of people who weren’t. And in conversation with them, a memorable comparison kept coming up, one that colored their skepticism about the promises of AI: Weren’t we equally excited about the BIM revolution? Why should we expect to be any less disappointed by what comes from the hype this time around?

We thought we were pretty funny.

I was early in my career when large design firms began transitioning from 2-dimensional drawings to BIM. The new technology was a topic of discussion in my graduate school classes, but at the start of my career my means of documenting (as with most others at the time) was in AutoCAD and other relic programs of the 20th century. The sea change on BIM started in the mid-2000’s, and as the chief editor for the local chapter of NCSEA, I teamed up with a few other young engineers to explore what it might do for our industry.

We spoke to architects, engineers, drafters and technology consultants. BIM had all the hype among industry professionals that AI does now, with a lot to be excited about: centralized data collection for easier storage and better coordination, improved design process quality, enhanced visualization, easier project management.

A better experience for the design professional, to be sure. But what was the overarching value proposition? Cost and time savings. This drove big companies to adopt BIM across projects, sometimes universally regardless of project size, project team composition or design complexity. Over time, modeling in BIM became a foundational skill for designers. At the same time medium and large firms created new departments around BIM as a core competency to contend with its increasing complexity.

Fifteen years on, it’s hard to argue against BIM technology’s positive impact on project outcomes. It has made possible the coordination of highly complex systems on large projects. And many studies have demonstrated that overall rework across most project types has decreased significantly. Some easy stats from Trimble (admittedly not a totally objective source) suggest that three quarters of AEC professionals report reduced rework on projects, and two thirds have seen improved budget and cost controls.

Why, then, did so many engineers approach me in Las Vegas with a bad taste in their mouth about the “BIM Revolution”? Many of them seem to feel that the benefits were felt asymmetrically - BIM reduced risk to the owner and lowered overall project cost, but designers saw comparably less commercial upside. Many I spoke with felt that they’ve been squeezed by an expectation that they can deliver the project at a lower overall cost while also delivering more detail and coordinating more between design disciplines and other project partners.

Without a doubt, BIM helps on complex projects. But it only works with tight coordination between partners, and that coordination is now a prerequisite regardless of project complexity.

The coordination requirements are worth some reflection. For BIM to really work on a project, partners need to agree on a lot - modeling standards, object metadata, software products and cloud platforms, naming conventions and more. The start of projects is loaded up with documentation to define the rules of the road for a project, and constant vigilance is required to guard against noncompliance. Even then, the design model is often rebuilt by the contractor, and rebuilt again by the owner at handoff. Today, 15 years into the BIM Revolution, things are better than they were, but there’s a lot of room for improvement.

Whether or not you think BIM lived up to the original hype, these cloudy perspectives offer us a cautionary tale we can take with us into the AI era. We should think about how we can develop robust, AI-powered workflows that can deliver value without project partners aligning first on a lot of details. AI is well suited to help us with this - it helps us structure unstructured data and translate data from one structure to another. This should make us inherently more tolerant to the nature and format of what we get from each other, provided we’re set up the right way.

Even if we can deploy AI technology in a way that doesn’t involve a lot of coordination and negotiation at the start of a project, we still need to navigate the change with a clear sense of what we want out of it. For design firms, adopting AI surely means seeking to improve the bottom line. To do that, it’s incumbent on our firms to both justify (or boost) the perceived value of their services and also reduce the operational costs required to deliver them. Doing that requires the participation of organizational leaders and marketing teams, as well as project managers and able-bodied innovators. It’s a whole-team problem.

I’m optimistic that we can define how our firms leverage AI to achieve our individual business objectives, and in a way that avoids the kind of complex coordination activities that came with BIM. By working together now, hopefully, we can set things up so we don’t need to work together later on tasks that don’t lead to a better design.