Early adopters report a spectrum of outcomes from AI technologies, ranging from success to unexpected challenges, according to a panel session during the 2023 National Institute of Building Sciences Building Innovation conference.
AI has been around for years, such as in smarthomes and smartphones. But as applications like ChatGPT, an AI-powered language model, or Midjourney, an AI-powered program that generates images, continue to capture the interest of the general public, construction executives are beginning to wonder how to use those tools in their operations.
Even the U.S. government recently expressed curiosity in AI use-cases, said Jay Kline, director of project management at NIBS, during the Sept. 6 panel in Washington, D.C. That’s a step in the right direction for the construction industry.
For example, the Federal Highway Administration recently announced a new $85 million grant program that state transportation departments can use to fund construction technology. That program will focus on digital construction methods such as computer modeling and 3D design.
“Construction is at the bottom of the digitization world,” said Kline. “Even though there has been a lot of progress, there is still a lot of business, growth and efficiency opportunities.”
Consigli, a Milford, Massachusetts-based general contractor, already recognizes those opportunities for AI adoption. The general contractor identified precision, performance and materials as three areas where AI is already providing impact, said Roger Grant, vice president of building technology at NIBS.
For example, with precision, Grant said ceiling plans are challenging for design teams. But a computer can generate a very accurate ceiling plan, or mechanical room space, he said.
“In buildings, there’s a lot of materials that go into a 20,000-square-foot building. There’s, for example, 3.6 kilometers of piping or cable trays, or 17 kilometers of pipe, and so we can use AI to reduce material usage,” said Grant. “The idea is that AI augments what the designer does to identify what the requirements are and then uses AI to optimize. That’s the main idea.”
Another simple use-case includes uploading PDF documents to a low-cost tool, like ChatGPT, in order to generate quick estimates, said Kimon Onuma, president at Onuma, a Pasadena, California-based architecture and planning company.
“We had a 100-page plan for a Tokyo project which included things like resiliency, risk, and what do we do in case of earthquakes, how much housing do we want, all that was written in it,” said Onuma. “We basically just fed it into ChatGPT and we asked it to summarize it for us. Then we used that to start doing our planning.”
Challenges remain for widespread adoption
A misconception of AI is that it is a quick plug-and-play type tool for construction executives. That expectation can leave professionals disappointed, said Onuma.
In reality, there are hundreds of AI tools available. For that reason, it’s important to experiment with these tools, in order to find the application that works best for individual projects and workflows.
Onuma compares this process to iPhone apps, for instance, where a user might download and then delete an email or music app in search for their preferred experience. A similar process of trial and error should be expected with AI too, said Onuma.
“It’s not perfect but it’s important to experiment and then use the lessons learned in your actual projects,” said Onuma. “That’s what we’re doing.”
AI also lacks the “human perspective,” added Kline, including true creativity and emotion. That’s why, although AI can help with code compliance or design plans, the tools still lack the creative touch humans possess.
“You can automate stuff today, there’s no reason to wait,” said Onuma. “But you must also be careful about what AI tells you because it does make mistakes, which means the eyeballs have to be on it.”