Construction is one of the least digitized industries, but jobsites are steadily transitioning from paper-based to digital workflows. Mining this new data will get more efficient with the help of artificial intelligence (AI). Watch the video to learn how AI has the potential to make construction smarter, safer, and more efficient.
Pat Keaney, Director of BIM 360 Enterprise Products, Autodesk: Ten years from now, I believe AI, machine learning, will be ubiquitous across all aspects of construction and all aspects of construction technology. I grew up with a single mom without a lot of resources. And when our back porch started disintegrating when I was 12, I looked at it, and I said, “Hey, mom, I think I can fix that.”
Manu Venugopal, Senior Product Manager, Autodesk: When I was kid, I was really fascinated with two things. One was motorbikes, and the other one was skyscrapers. Eventually, those buildings won.
Shubham Goel, Data Science Manager, Autodesk: My grandfather’s in construction. My father spent his entire life in construction. Now, I can claim that I’m a third-generation construction person at family reunions. Although every time I share the story, they say, “Unless you have calluses on your hand, you’re not in construction.”
Over the last 10, 15 years, construction jobsites have been going through a transformation. They have been going from a mostly paper-based workflow to a digital workflow now.
Venugopal: This has resulted in an explosion of data sets, and that’s an opportunity for leveraging machine learning, because nobody has the time or energy to sift through all this data.
Goel: Most of the industry right now has data that’s not as well organized, and it cannot be easily consumed for analysis, and it’s really hard to extract meaningful insights from that data. And that’s really where machine learning can go in and help them.
Keaney: Machine learning is a technique within the broader umbrella of artificial intelligence. For example, if our customers have a thousand quality issues on a construction jobsite, no human being could or would want to read them every single day.
Goel: Instead of hiring a team of people that go and look at their data for the last 10 years and categorize and clean it, they can have a machine-learning model that looks at that data and that gives them meaningful information out of that in seconds.
Venugopal: I think of this as a way for computer algorithms to learn from data and start making recommendations. And once they’re trained, they can start predicting things based on what they’ve learned.
Keaney: What’s really interesting about machine learning, to me, is the application of it.
Goel: In the past, folks have had to fill out long forms whenever they want to capture any information on the jobsite. We have seen checklists with more than 150 items. And machine learning can drastically simplify this process. In the future, people can just take a photograph, and from that photograph . . . As it’s said, a picture is worth a thousand words. It’s also worth a thousand insights, and all this information can be automatically extracted, and that long checklist can be automatically filled out.
Keaney: There’s a lot of knowledge in the construction industry. The problem is most of that knowledge is locked up in the head of a superintendent who’s been doing it for 30 or 40 years.
Goel: But when they leave or retire, all that knowledge goes away with them. And now, how do you build that knowledge set again? And machine learning is something that can help or augment that experience because it can learn from the data.
Keaney: If we can capture that knowledge and turn it into an assistive app that you can hand over to a young, bright, college-educated person who’s interested in construction, you can help them be more effective. So, you can set up the industry for the next generation, and you can also address the labor shortage.
Choosing the right subcontractors for the right job and finding that fit really matters. Prequalification is often done purely on the basis of financial analysis: Is this company solvent? Do they have the correct amount of insurance? Et cetera. So, what if we could combine your historical knowledge of those subcontractors and the actual work they did on the project with the financial analysis to help make sure you get the right subcontractor for the right job?
Goel: That is some place where AI and machine learning can tremendously help.
Keaney: You can have a great electrical subcontractor, but maybe they’re more experienced in one project type than another. All that information, we think, will also help upstream in procurement.
Venugopal: It’s not just about improving the quality of a construction project. It’s also about making sure we can identify safety problems, safety hazards on jobsites.
Goel: One of the things that really surprised me when we started looking at construction safety was the number of fatalities in this industry. In 2017, there were 971 fatalities in construction in the US alone. OSHA, the US body for health and safety, believes that more than 60% of those fatalities are completely preventable.
Keaney: The world that I’d like to see is that we take the guesswork out of a lot of this, and we just surface what you need to know, when you need to know it, and where you need to know it. So, sensors that are monitoring jobsites, image rec that can detect a problem and point it out to you. If you’re a superintendent, the AI can say, “Hey, go look at this. There’s a welding issue downstairs that’s creating a safety problem.” And that can get automatically detected, automatically send a notification out to everybody on the team.
Goel: Being able to understand risk before something really bad happens on a jobsite can prevent some of those injuries and fatalities and keep the folks on the jobsite safer.
Keaney: If we saved one life, it was worth it. If we keep one GC or subcontractor out of court, it was worth it.
Venugopal: I truly want the next generation to think of construction as a really exciting place to work. I have a young daughter, and I want her to be excited about construction 10 years from now. This is where we see a great, great opportunity where we can help change an industry, and how things are done, and make it really a better place.
Keaney: What we’re doing with AI and construction, I think, is really interesting. But the number of problems that can be solved by us, by partners, by the ecosystem, I really believe it’s unbounded.