A McKinsey & Co. study released earlier this year predicted that the engineering and construction sector will be slow to embrace artificial intelligence (AI), but despite this slow adoption, presenters during a webinar hosted by Engineering News-Record said it can help E&C companies expedite early processes, create the best plans for projects and identify if a project is starting to go awry.
Rob Otani, chief technology officer at structural engineering consulting firm Thornton Tomasetti Inc., explained that AI is already all around us — for example, look at Gmail's suggested replies, which are inputted automatically without the users' involvement as part of a pattern-learning algorithm meant to save the user time.
The key is finding where to leverage AI in professional capacities, Otani said. When designing structures, for example, AI can automate mundane and repetitive tasks, thereby allowing engineers to spend more time creatively solving problems, he continued.
AI's ability to process and analyze millions of data points also means it can understand patterns and even detect ones a human can't, he added.
Thornton Tomasetti, which has been studying machine learning for three years, developed a software application called Asterisk that it claims performs structural design of a building in seconds. The app uses the company's past building design data and machine learning algorithms to choose structural member sizing for steel and concrete building elements, including floor framing, columns, slabs and foundations. Only a building massing has to be manually inputted.
Other AI benefits
In addition to design benefits, AI tools can learn how to evaluate a project from a financial standpoint and detect if projects are starting to go over budget. Although Otani wouldn't recommend using AI for final design decisions, he applauds its benefits for initial design considerations, saying that 20% to 30% of engineering workflows could be automated using AI.
During the webinar, Andy Leek, vice president of technology and innovation at Paric Corp., also said AI is particularly beneficial in the early design stages so the team can run through myriad design options quickly. "AI can make processes ... take a whole lot less time," he said, adding information machine learning also ensures decisions are data-driven.
Leek advised E&C companies looking to leverage AI to research the tools they already use, define early achievable goals that make sense for the business and then network with organizations that could help reach those goals. Companies don’t need to create new tools; rather, they just build a network and leverage existing technology and relationships to find the optimum solution.
Iro Armeni, Ph.D. candidate at Stanford University, pointed to the construction industry’s much-publicized lack of productivity. Since 1995, the industry has seen an annual productivity decline of more than 1%. Adopting new technology and management techniques could increase its value by $1.6 billion, according to a McKinsey Global Institute report.
One challenge the industry faces, Armeni said, is a lack of understanding about what AI can do for construction. If used correctly, it’s a powerful tool, she said.
ALICE, a tool Stanford developed, can explore how to build a project taking any variation of zoning, sequencing, resources and equipment into account, Armeni said. While it would take a human more than 13 years to calculate all of the available combinations, he said the tool can work through processes related to planning, scheduling and management on demand.
Meanwhile, TechCrunch reported recently that UpCodes AI is a software plug-in that uses AI to run a "spell check" on a BIM models' compliance by flagging violations. The product, still in beta testing, identifies an average of 27 code violations per project, according to CEO Scott Reynolds.
AI isn't restricted to large companies. But the presenters said that small firms should keep two challenges in mind. One is obtaining and harnessing a lot of data. The other relates to programming, but users can hire consultants to do just that portion that if it isn't feasible to do it in-house, they said.
Although expensive, AI becomes part of the return-on-investment equation, presenters said. If machine learning could save twice as much as it would cost to hire a consultant, it could be worth the investment.