Fifty years ago, my father and a few fellow farmers had a vision to reenact their childhoods. They grew up witnessing a huge agricultural transformation in the Midwest. The tools and technology they used working on the farm drastically changed as they became adults.
In an effort to capture the nostalgia of their childhoods and relive America’s industrial revolution, my father and his friends spearheaded an annual exhibition to demonstrate the technology spanning the last two centuries. This past August, my hometown celebrated the 50-year anniversary of this event in southwest Minnesota.
For the unfamiliar city dweller, this exhibition is quite impressive. The technology displayed is vast and highly specialized, from wooden threshing machines from the 1860s to steam powered tractors from the early 1900s. Most of the brands on display do not exist today, as the technology has been consolidated into a handful of major players. When watching the history of this renaissance period on Tractor Wars (Iowa PBS, 2024), farmers achieved 100x productivity gains over several decades by augmenting horse/human power with machine power.
Today, we are witnessing a similar technology revolution driven by AI but at a dramatically faster pace. I heard this first-hand at Autodesk University this fall where AI was front and center. Many thought leaders I spoke to believe the AEC industry will experience a massive transformation over the next five years.
In a way, the timing could not be better. AI has great promise to transform our fragmented, labor-intensive construction industry, which struggles to be productive (see the chart below). The only way that I can see us building more, faster, and with less labor and material is by increasing productivity.
One of the barriers for productivity growth is the skilled labor gap and inability to transfer knowledge fast enough across generations. By combining AI with human intelligence, it will essentially codify human intelligence to accelerate generational knowledge transfer. Just as farm machinery like the grain binders and harvesters codified human labor, machine intelligence will encode company knowledge and the individuals. One of our industry’s systemic challenges is attracting and retaining a workforce. Many domain experts will retire in the coming decades. Using AI to codify this knowledge so we have less “friction loss” will make our industry more productive over time.
A good example of this codification is a recent interview with Rob Otani, CTO at Thornton Tomasetti. (Podcast, Domo Arigatou, "Mike 2.0 " UnfrozenJan 29, 2024). He talks about how they codified the expertise of an engineer called Mike who was their guru on welding inspections. Anything about that topic, Mike knew the answer. Using Mike’s emails as an AI data training set, they were able to reproduce his responses to a high degree of accuracy. This is a way for engineering firms to continue leveraging their firm's “best” IP over time, which indirectly helps the industry advance as well.
How should we prepare as an industry to best leverage the potential of human-machine intelligence? I see three key tenets to best shape this new landscape:
As I reflect on past technology disruptions, I see evidence of these principles and believe they will apply to AI. Let’s explore them more in depth, and why I think they are relevant.
Just as machine power converged many manual activities a century ago, I expect machine intelligence to converge human tasks in many fields including design. AEC businesses and its professionals will need to rethink how they specialize or differentiate their services as AI converges their workstreams. Using an earlier example from agriculture, the leader John Deere was initially very specialized in the market focused on making plows. However, when steam power fueled new markets, they decided to expand to making tractors as convergence was where the future was headed. In AEC, we should consider if steel, concrete, and timber design technologies will converge like wheat, corn and soybean harvesting technologies did in farming. Will AI’s ability to orchestrate highly specialized agentics converge the tasks of engineers, detailers and builders into one master builder?
There is great potential here to converge workflows across our industry. But how will humans and machines interact with each other to maximize the impact on AEC?
A decade before my father started demonstrating the use of machine power in agriculture, an engineer working in the US Navy was thinking about the future of machine-human intelligence. Douglas C. Engelbart's seminal 1962 paper, "Augmenting Human Intellect: A Conceptual Framework," provides a remarkably prescient and foundational perspective on how AI could be applied in fields like building design and construction. His central thesis was that technology should be used to augment human capabilities, not replace them. He envisioned a system where humans and computers work together, with the computer serving as a powerful tool to extend human cognitive and creative abilities.
This augmented approach makes sense to me as I reflect on a favorite 1980s TV show called Knight Rider. KITT was the cool AI that assisted the risk-taking Michael Knight on his quests. KITT assisted Michael to work as an effective pair at solving crimes. KITT could never replace Michael, but he most definitely gave his human partner a superpower and made AI look indispensable.
Engelbart's core argument against the idea of technology replacing humans is the most enduring part of his framework. He defines "augmenting human intellect" as "increasing the capability of a man to approach a complex problem situation, to gain comprehension to suit his particular needs, and to derive solutions to problems." He does not see the computer as an autonomous intelligence but as an "artifact" and a "means" that, when combined with human "language" and "methodology," forms a powerful human-computer system.
How integrated future intelligence machines will shape our future is yet to be discovered, but I have no doubt that humans and machines will become more and more interconnected.
As observed in the evolution of any system, progress is not a simple, linear process of "more data," but a synergistic combination of learning, exploration, and fundamental principles. This was evident during the industrial revolution, and it will become evident once again during the AI revolution.
My central claim is that progress in a complex system like intelligence is not solely dependent on learning from past data or experience. Instead, it's a dynamic interplay between 3 types of knowledge:
I believe the most powerful human-machine intelligence will blend components of large-scale data models (for experience and efficiency), exploratory algorithms (for creativity and breakthroughs), and rule-based or symbolic systems (for logical grounding in science and safety). These facets have been the cornerstone of human and technological innovation for centuries and will continue in the digital age of AI.
The relationship between machine systems and engineers is certain to evolve within the construction industry as the AI revolution grows more and more powerful. Being a member of Structural Engineering Institute’s Optimal Design Committee back in the 2000s, I had the privilege to meet a few of the trailblazers in the structural optimization space and see real world applications of evolutionary computation for topology optimizations (Shook, 2009, CTBUH). Today, there is compelling research for how to effectively implement structural optimization concepts where humans are still in the driver's seat (Mueller, MIT, 2014). Structural engineers don’t need to reinvent the wheel here but build upon the decades of proven use cases.
It’s an exciting time to witness this transformation across all parts of our lives, not just AEC. In a recent NCSEA AI Strategy team’s Summer Series, Kris Dane (Thornton Tomasetti) and MZ Naser, PhD (Clemson University) talk about why this disruption is different from past disruptions like CAD, BIM or even digitalization. AI systems will be ubiquitous across most aspects of life – personal and professional. We will need to determine as specialized industries how we integrate and use these different AI systems.
After recently attending Autodesk University, I see that the adoption is already happening and could have a notable impact within five years. However, I expect the structural engineering industry will adapt to AI slowly and deliberately. We saw this with equally profound innovations like finite element analysis (Liu, Shaofan, Park, 2022, “Eighty Years of the Finite Element Method: Birth, Evolution”). If, as an industry, we want to manage AI’s advancement more quickly, we should do so in a strategic way. In his book AugmentIT, AI visionary, Mehdi Nourbakhsh, challenges businesses to ask tough questions about not just what we do with AI but how we do it and why.
Whether it will take a few years or a decade, AI is here to stay. We are in this together and need to reflect on what we want our future world to look like after the AI transformation. Historical exhibitions, like the one my father founded, are great ways to learn about industry consolidation and transformation. I expect that in a 100 years, we will have similar reflections about AI and intelligence systems in structural engineering. What do you think future engineering generations will say?
Michael Gustafson is a seasoned business strategist in the AEC tech sector with a focus on structural engineering and fabrication. He practiced as a structural engineer at Ellerbe Becket, holds an MS in Civil Engineering, an MBA from Michael J. Coles College of Business and is a Professional Engineer from California. He is also certified in AI for Business Managers from MIT. Michael is currently Vice President of Strategy and Business Development with Qnect, a service provider of data insights and efficiencies who is unlocking new and sustained value in the construction industry. You can find him on LinkedIn at: https://www.linkedin.com/in/michael-gustafson-2047786/.
Have comments or feedback on this article? Visit its AECbytes blog posting to share them with other readers or see what others have to say.
AECbytes content should not be reproduced on any other website, blog, print publication, or newsletter without permission.
