The Changing Attitude Toward A.I. in Construction: Are We Finally Ready?
By Adam Handfinger and Quentin Davis
As we move into 2024, we are experiencing a seismic shift in machine learning capabilities that power AI, and it’s still the tip of the iceberg. As we build through this shift, what do construction companies need to know, and how much of the industry will be left behind?
How big is this shift?
Ten years ago, AI was one of those technologies you only whispered about in construction. It was hard to tell who was for it or against it. It sparked fears of unemployment, data manipulation, and the death of hard-built, long-held industry relationships. Five years ago, you could talk about it more openly, but it was mostly just talk. Adoption and implementation were low and slow. At that time, you may remember using Google (yes, also powered by AI) and adjusting your search word by word to get the answers you wanted. Google needed to understand what you literally meant, not what you may conceptually want. AI could handle simple questions and simple contextual information.
AI development continued, but if you weren’t a developer, you probably weren’t paying attention. Then two years ago, ChatGPT showed the world what AI was finally capable of and made it impossibly easy to use, democratizing AI for everyone. While it wasn’t the first application to harness large language models (LLM) at dizzying speeds, it was the first LLM whose capabilities excited the masses so much that ChatGPT became one of the fastest growing products in history. ChatGPT showcased an LLM that could grasp lengthy text and unstructured data and generate responses that could read between the lines and be accurate, coherent, and contextually relevant. Previous LLMs returned babbled responses to questions; today they answer with insight that rivals expert human analysis.
To put it in perspective for our legal friends, LLMs two years ago could barely understand the bar exam. LLMs today can pass it without struggle.
As these developments in AI models were happening at a macro level, we finally began to see the mindset shift we’d hoped for in industries like construction. Seeing the power of AI, and maybe even testing it out through ChatGPT, contractors realized they should not fear using AI, they should fear not using AI while their competitors do. Work ethic, specialized labor, and industry relationships alone can’t keep up with the blurring pace of this technology.
AI isn’t a new language, it’s all languages
Our Principle Software Architect, Quentin Davis, said it best, “Right now, we are living through a seismic shift in what is possible in AI. The gap between the models a few years ago and today’s models is the Grand Canyon of cognitive capabilities. Until just a couple of years ago, the available AI models were like a tourist who looks up everything in a phrasebook but fails to grasp the foreign language around them. Today's AI models aren't tourists, they speak, understand, and write the language like locals.”
“Today's AI models aren't tourists, they speak, understand, and write the language like locals.”
He went on to explain that language is more than just text. It’s more than a mere medium of information from device to person to device. “Language is a fundamental primitive of cognition,” says Davis. It creates meaning, understanding, culture, emotion and so much more. So the more capable an AI model’s language abilities, the faster it learns and the more it can automate for humans. As LLMs advance, they can do so much more than input, output, and analyze. They can converse, predict, advise, argue, and even joke.
From “What is AI?” to “I need it now!”
Just a couple years ago, Document Crunch’s team spoke at industry conferences and fielded questions like “What exactly is AI?” and “Why should we use it?” Today, those questions have changed dramatically.
According to a recent global survey, 45% of people working in the AEC sector globally say they currently use AI at work. We’re approaching a tipping point where a majority of the industry will be using AI. As that happens, contractors realize fearing AI is the risky part, and in some cases negligent. They should instead fear what happens if their competitors are using AI. A project management firm recently studied 64 building sites around the world and found that progress varied by 30% week-to-week even at the most productive sites. Imagine a firm with AI tools capable of solving those inefficiencies. That firm’s competitive advantage could be using powerful AI technology.
General contractors now realize that AI can be adopted quickly and trusted to make them more efficient. They no longer need convincing, they just need guidance to pick the right tools.
The shift from “Should we trust that AI is working in the background?” to “Can you tell me all about your AI so I can use it to improve my business and brag about it to our shareholders?” is finally happening, and it’s a technology provider’s responsibility to deliver on the promise of AI.
AI Will Set a New Construction Project Standard
AI is leveling the playing field for information. Right now, builders may win projects because they have the best estimates and the best relationships. The availability of data through AI will change that. Preconstruction, risk, legal, and project management teams have to get smarter and focus on thinking deeper, not just reporting information that AI can find and analyze. On the construction site, for example, AI will allow us to know exactly what work was done each day (percentage of drywall complete, utilities installed, etc.) and get real-time factual answers to requests for information. Bid and project information will be fully democratized, specific, and available. Contractors will need AI software to play at the same level, then they’ll need to add value and expertise to win.
Consider this vision from our Co-Founder and preeminent construction attorney, Adam Handfinger, “Think about the efficiencies you get from building something in a factory environment. Items are made in mass, all following the same process, reducing material, labor, and operations costs. Imagine if construction was more like that. Imagine you have all the information you need to build a complex structure in a controlled, frictionless environment.” While we don’t actually expect all construction to happen this way, we can imagine the project volume and profit margins possible if a fraction of projects did. We can imagine the time and mental capacity this would free up to allow project teams to add value and uniqueness to the most critical projects.
So, Who Will Get Left Behind?
Hopefully, no one. Realistically, companies who fail to consider the following during this monumental shift will struggle to compete five years from now:
The best AI learns from inputs over time. The best language model you can use, the one most tailored to your company, is the one you’ve been using the longest. As more domain and company knowledge is collected, the better it becomes.
AI will produce data and analysis that will impact your whole business. You’ll see the full picture of your business clearer than ever before, possibly requiring more change than you anticipated. The sooner you get this clarity, the sooner you can act on it.
AI still has a learning curve for users. Not because it’s incredibly complicated or tedious to use, but because it can be a completely different experience than using legacy software applications. Construction Dive wrote about how reverse mentoring can help address this, as will early adoption and giving employees time to learn and adjust.
The construction market is competitive. Supply chain uncertainties and labor shortages make it hard and unpredictable to compete on bids alone. Contractors will need new ways to add competitive advantage. Technology, through AI, will provide that.