Q&A: How AI could change the construction process
Artificial intelligence expert Jeevan Kalanthi offers his advice for tech-curious builders.
Jeevan Kalanithi, CEO and co-founder of OpenSpace AI, has seen artificial intelligence grow from a small, emerging field of technology into a powerhouse set to disrupt countless global industries. We spoke with the San Francisco-based tech entrepreneur about how he has seen the field of AI change and what areas of construction it could disrupt.
Be sure to catch Kalanithi and other industry experts in Vancouver, B.C. for the 2023 Independent Contractors and Businesses Association (ICBA) Construction Innovation Summit on Oct. 30th and 31st.
SiteNews: What is OpenSpace?
Jeevan Kalanithi: OpenSpace exists to simplify the lives of builders and what we do is pretty straightforward. Our core capture product makes it really easy to have a full visual record of any space, indoors or out. Think of it like Google Street View. You can image it as much as you want to, every day or every week, so you can see what is there without having to physically be there. And you can see what was there yesterday, a week ago, two weeks ago, five years ago. And the point of it is very simple. We want to create a record of what is actually happening on a job site. You don’t have to wonder what is there and you can replace the opinions, memories and really laborious workflows. Think of any RFI or change order issue, if you could just look and see what’s there you could resolve it in seconds instead of with a mountain of paperwork. What we built on top of it are AI-powered tools to understand what is actually in that imagery and reality data, and then classify and quantify it. We call it OpenSpace Track and with it can do things like tell you how much drywall you’ve hung versus taped versus framing and compute the percent complete and tell you if you are ahead of schedule or behind. You can start to really answer those productivity questions that are really at the heart of what builders do.
How did you get involved with AI?
AI was something I’d been interested in since college really. I did a degree called symbolic systems that was unique to where I went to undergrad. Think of it as a very technical cognitive science major. I did two concentrations, on in philosophy and the other was artificial intelligence. Then I went to graduate school and further pursued studies in artificial intelligence. So I have been doing AI stuff academically since college and, entrepreneurially, I have been applying these techniques to buildings since before OpenSpace.
What are some misconceptions people have about AI?
First, AI is not just one thing. The way we think about it today is this text-based, generative AI system where you put text in and you get a bunch of text out or an image or something. But AI was a term coined in the 50s. Back in undergrad I took a class from a guy named John McCarthy, one of the “fathers” of AI. And in grad school I had one with Marvin Minsky, the other “father” of AI. It was pretty cool to see what those guys were working on. It was totally different stuff actually than what you see today. If you even go back and think about robotics and autonomy, that’s AI too. And it’s what has powered self-driving cars. Here in San Francisco, they work and I routinely take them around town. But that’s pretty different than generative AI. Even further back to the systems that beat Garry Kasparov at chess. Those were a totally different flavour of AI. It’s not just one thing and the different sub-branches will that different applications for builders. The second misconception is that these systems think like human beings do. They don’t really. The latest AI systems don’t really. One way of thinking about it is that they don’t know what they are talking about. They give very knowledgeable and cogent responses but it’s not clear that they have a deeper conceptual understanding of what they are talking about. The third misconception is that we have a good understanding of how these technologies will influence how we live and work. We don’t. These things are going to evolve in ways that are shocking and surprising and that we can’t even think of today.
How might these emerging technologies change how we build things?
I think architecture is going to change a lot and engineering too with generative design. These systems are getting really good at taking plainly spoken parameters and turning them into designs. So these armies of junior architects doing detail work, that might go away. It would be great. It would empower architects to do the more creative design work that we actually want them to be doing. I also think there will be applications for robotics on the job site to help alleviate labour shortages, but I don’t think it’s going to be “Terminator” robots building the way people are building. They will just look like tools that are a bit more autonomous. Think about the evolution from hammer to nail gun. That was a big deal. Using a hammer is pretty annoying if you are used to a nail gun. I think these robotic solutions will feel that way. It’s not going to be doing general purpose things, it will be specific. I also think the ability to have a clear, indisputable record of the project to make decisions is going to get a lot better. Computer vision and AI is going to allow us to not need to go to these job sites and make reports. We can have the answer at our fingertips no matter where we are. I think this will really change how labour is distributed. And by labour I mean both white collar jobs and people actually doing the building. Lastly, and I think this will be pretty transformative, I think the amount of paperwork has a chance to go down significantly. Because so much of the reporting and paperwork in construction is based on trying to have an accurate record of what’s there. Think of RFIs as an example of that. AI has the opportunity to put real data at people’s fingertips and get rid of a lot of that unnecessary paper pushing. I think that will change a lot of roles for people who have construction management degrees. They will be able to spend more time out in the field helping get things built. It may even help us contract differently. There could be less need to divy up the risk into buckets. You could have an AI-powered transparency layer that can more effectively allow people to prove their work and share the risk.
How has AI changed and evolved over the years?
When I was studying AI in college, some of its true pioneers were there. It really was the cutting edge at that time. But AI at that time was very much about rules-based logic systems, like a system that can do logical proofs was what people thought of as being AI. That changed a lot once I got to grad school. The focus was really on machine learning or statistical methods. It’s just about recognizing patterns. The math of that was different and it was also enabled by the amount of data generated by the internet. So you could actually start training these pattern recognition systems in a way that wasn’t practical in the 80s and 90s. I remember my first neural network I built as an undergrad. It sounds amazing but it was just a classroom assignment. It was able to recognize pictures of letters and classify them as an “A” a “B” or a “C” and so on and classify it. The level of sophistication for that vs. what I did in grad school vs. what we are doing now is almost the difference between a single celled organism and a tiger. The amount of change is absolutely insane and I would say that the biggest change between grad school and now is the amount of computing power we can dedicate to AI problems now would have been ridiculous and impractical even 10 or 15 years ago. So a lot of the methods and math that people knew about back then was just theoretical. But now you can actually build a system that does that. The amount of data these systems can consume and the amount of computing power we can devote to them was unimaginable even a few years ago. That’s creating huge unlocks where you can create systems with unbelievable sophistication that wouldn’t have been practical even two or three years ago.
How can builders prepare for the changes AI could bring?
Focus on your business and the actual problems you want to solve. There is so much BS and snake oil out there, which is true of so many industries, including technology ones. Don’t be afraid to focus on your issues and don’t worry too much about missing the boat. Second, pay attention to the more tech-interested folks on your team and see what they are messing around with. That doesn’t mean they will be right about everything but they can be your antennae to hear what’s out there. You don’t need to be going to computer vision conferences yourself. Reading is good. Subscribe to the MIT Technology review. It is written in plain english and has great articles in it. It gives you a sense of what is going on. Thirdly, don’t be afraid to experiment with new things, especially things your team is bringing to you. Give it a shot and see if it actually helps you. Lastly, see what your competitors are doing. You don’t need to be an expert in AI. You are an expert builder. That is what you should focus on and why you are awesome for society. You don’t need to try everything, just things that help you run your business more effectively.
Get tickets to see Kalanithi and other cutting edge construction leaders at the ICBA Construction Innovation Summit here.