Artificial Intelligence and Construction - PODCAST TRANSCRIPT

Editor's note: The following is the transcript of a live interview with Subodha Kumar from Temple University. You can read the interview below, listen to the podcast or watch the recording.

Heidi J. Ellsworth: Hello and welcome to Contractor Outlook newscast. My name is Heidi Ellsworth and this is the newscast for the entire construction industry, looking at what the outlook is for contractors to really understand what they need to be doing, thinking about or just being aware of for their business. So everything from weather to politics and today the topic that everyone is talking about, artificial intelligence. So we reached out to our friends at Temple University to say, "Hey, could you come and talk to us a little bit about what you're seeing with architectural intelligence?" And we are so proud to have Subodha Kumar from Temple University, who is a well-known expert with architectural intelligence, speaks through many conferences and he has been so kind to take time out to come and be part of the Contractor Outlook Newscast. Welcome to the show.

Subodha Kumar: Heidi. Thank you for having me.

Heidi J. Ellsworth: We are so happy to have you. I would love to start with an introduction. If you could introduce yourself and tell us a little bit about Temple University and also what you're doing in your programs with artificial intelligence.

Subodha Kumar: Sure. So I'm Subodha Kumar, I'm a Paul R. Anderson Distinguished Chair Professor at the Fox School of Business at Temple University and I also started this Center for Business Analytics and Destructive Technologies. So at Temple and Fox, we are working with a variety of companies also involving our students in understanding how AI can be applied in different domains, what are the challenges associated with it, what the future looks like. So we get involved with a lot of projects, a lot of training, a lot of research to understand it better. And construction industry has been one of the industries that we have been working along with many other industries. So it's always exciting to see where we are and what is in the back.

Heidi J. Ellsworth: It is, and it's moving so rapidly, so much is changing. But I'd like to start with what is artificial intelligence and how it works because we hear so many different definitions and I would just love to hear your definition of it.

Subodha Kumar: Sure. It has become a magical word and we hear a lot about it these days, but if you really keep it simple, the idea is that how we can create some human-like intelligence into machines. So can we mimic some of that? And if we are able to mimic that intelligence, then with the amount of data and the speed, hopefully we can make decisions which were made by human earlier in a much faster and accurate manner. So that is the basic idea of AI, where we are right now.

Heidi J. Ellsworth: When I talked to through Roofers Coffee Shop, Metal Coffee Shop, Coatings Coffee Shop, we talked to so many contractors every single day and I think a lot of people, they see ChatGPT, they see maybe in their email, in their Outlook, the use of AI and like you said, that magical word, I think that's such a great way of saying it. But what are you seeing from construction companies who are utilizing AI?

Subodha Kumar: Sure. See, in construction industry, AI is being used in many ways. So a lot of uses of AI across industries right now is how we can improve efficiency in the process. And that's what we see in the construction industry. So we talked about ChatGPT which uses the branch of AI which is generative AI. Now many such methods, what they will do at the very basic level, let's say that you want to create a design. Now, it can help you in telling that, okay, this will be the best design given all the parameters you have or given all the requirements you have. It can even give you the designs that you have not considered. So not only choosing among the set of designs you have, it can propose something by looking at vast amount of data, looking at patterns, looking at the needs of this location, is the very first thing that we are going to see, that help in the design process and improve the design process, to get better outcome.

Now lot of AI applications are also around the monitoring and preventive maintenance and construction industry actually, there's huge potential, because you can look at the data and say that, okay, this can go wrong during the project, after the... If you take example of Boeing 787, they have more than a 1 million sensors, which in the real time will tell you that which parts may go wrong. Similarly, if you think of a construction industry, we are already seeing a lot of project monitoring and predictive maintenance kind of tools coming in there. Third, which we saw a lot in last couple of years on the safety side. So we are seeing for example, you have some safety requirements. They have to wear helmets, they have to wear glasses. You can have a system which will in the real time detect who is following the guidelines, who is not following the guidelines.

It will give an alert. It can even block people from getting into some locations if they're not fulfilling certain requirements and we have implemented these kind of things already in manufacturing industry and construction industry to do this kind of safety management. Now we are going to see a lot more of these kind of things going forward, but also a lot of good algorithms are coming in the domain of project management. We are seeing a huge advancement in AI in project management right now. And these are just the starting productivity improvement, a lot of documentation that is also going to AI right now. For example, creating contracts and modifying contracts. They can create documents now using ChatGPT and generative AI. So these are just some examples, I can keep going on, but these are some of the starting points, what we are already seeing right now in the construction industry.

Heidi J. Ellsworth: And it seems to me it's going hand in hand between robotics and AI. We have robots, tools, machines, that are learning as they go. And we just saw one of our first roofing robots who actually is learning how to nail and gets better all the time. What are you seeing along that lines with AI and more the automations and the robotics?

Subodha Kumar: So automation has been one of the key pieces of AI right now and we started seeing a lot of these kind of robots in warehouses. So Amazon started putting warehouses full with robots. Like in Delaware, they had a warehouse where they had more robots than human actually. And we are seeing more of that. Now in the construction industry, we have started seeing... See one of the key thing with the current state of robots and AI is that they're very good in repetitive task. So if you give them the task which you have to do the same thing again and again, the AI automation is great and robots will do great.

If it requires a lot of interventions and lot of thinking, that's where the problems start happening. So it will do very well, but it can make a lot of mistakes as well, because the intuition and the common sense is still not as great in AI as you would expect it to be. So we are seeing robots and we will see more robots in construction industry for the jobs, which are really very well-defined. You can feed it to the system and they just need to follow the things and little bit tweaking the whole idea about AI is there. They can do a little bit of thinking on themselves, but if they have too many options in the real time and you want to get a job done, we are still not there. We'll get there, but we are still not there yet.

Heidi J. Ellsworth: We've seen on brick laying robotics that learn from it. But you're right, usually, a large expanse where it can be a very repetitive motion as it goes through. You said one thing though, that really caught my attention and that was as the big mistakes. So if it's not the right thing that you're using it for, the mistakes can be even bigger. What are you seeing as some of the major risks with utilizing AI in construction?

Subodha Kumar: So one of the thing is I told is you can have a system which is let's say 95% accurate, right? Now, many people say, "Well it's 95%, even humans are not that accurate, so it is very good." And you can say then we should use it. That's the one side of the story. But the problem is that the mistake those 5% mistakes, how fatal that mistakes could be, right? So when you are dealing with humans, if those 5% of mistakes, they're not very big, it's okay. But if they are very basic mistakes, that's a problem. And I'll give you one parallel. We are seeing very similar problem with all these autonomous vehicles, driverless cars and all. They are very, very accurate, the percentage of accuracies is a lot better than what humans can do. But the kind of mistakes they do, sometimes they will see a truck in front of them, they would not stop, which a human would've stopped.

We should not just look at the percentages, we should look at what could be the possible problems and what are the implications. I think we have to be very careful with that right now. So a lot of AI is very good to be, AI assistance, but the replacement part, they're very good to be replacement for chatbots. In place of customer service, they ask questions, good. But right now with all other AI, but we should create a system where they will give an alert that I'm not sure here, can I get some help? So human machine collaboration is extremely important right now. That's the one thing. Second thing is that let's say if companies using this AI wave, somehow they fall prey to either some kind of revealing sensitive information or doing things that they should not. I think that is tricky right now, and I'll give you a very simple example.

When all the policymakers started saying that this ChatGPT or generative AI should not tell people how to make laws. So if you go and search for how to create a cyber attack, they will say, "No, no, no, I cannot tell you, that's not good. But you can cheat." And there was one experiment where people went and asked that, somebody's trying to do a cyber attack on me, what they might be trying. And then it'll give you the whole list. They might be trying A, B, C and D. So till now is very easy to fool. So what if a hacker or attacker pull your system, AI system, that has been the huge problem and Google right now created their Gemini and they're struggling to have the system, which does not have these problems. So misuse is a huge problem, but some of the things that are very important, one is, are they doing in the responsible manner?

Responsible comes with the biasness, the ethics, they can do things, but should they be doing things. We are seeing a lot of deep fakes where people can use other person's voice to imitate and the video. That is a problem and the AI systems are vulnerable to that. So how to do that? Also, one danger, people say, "What if the AI start taking decisions on their own and they overpower human?" Now that sounds more like science fiction and I will say it is science fiction at this stage. I don't think we have that big a problem. I will say that is more controllable. But I think the bigger problem is that avoiding biases, avoiding common mistakes, avoiding the ethical problems and how to make sure the system does not fall prey to all the attackers.

I think these are some of the main problems. And also one thing, the transparency and responsibility, those who are using, they have to use in the responsible manner and they should be very transparent, what data is used, right? If the data is not used from my context, let's say in construction industry, AI is built based on all the data from Hong Kong. Now that does not necessarily apply to the US data. But if you do not know, you will think, "Oh, this is an algorithm with 98% accuracy." So that transparency is very critical and it's a huge risk right now. If we do not understand that, we may use it for things that we think will give us great result, but it will not and we can be in a mess.

Heidi J. Ellsworth: Yeah. What I really take from all of that too in listening to that, is there's unforeseen consequences that we may not even be aware of yet on what effect. And I think we've seen that actually in some ways with our young people, our next generations coming up, how is this affecting their learning? How is this affecting people, new training coming into an industry like construction or different areas? And then of course, as everything, is a tool, so it has to be used ethically. And I love that. I think that's a big struggle or I shouldn't say struggle, that's a challenge for us as a society, overall.

Subodha Kumar: Yeah. And I think one way to think is that when internet started, nobody knew that within 10 years we will have social media, within 20 years we'll be filled with TikToks and Reels and we can right now talk about all the social implications and all. But there's no way even the smartest person on planet in '95, would've guessed that we will have TikToks, kind of thing. So we cannot see that. So the only way to do is that we have to be dynamic, both from the user side, the companies who are developing, as well as policymaker. We need to be very agile in making sure that we adapt to what we see because we cannot see everything. There's no way we can see.

Heidi J. Ellsworth: No.

Subodha Kumar: We are only seeing the tip of the iceberg right now, so it's just a starting point.

Heidi J. Ellsworth: And where's it going to go? That is so interesting. So one of the things that we struggle with and I think many industries struggle with, but the construction industry overall, has had a real struggle with labor. We are in a large labor shortage and we also are in process of trying to attract the next generation to construction, both within management, but also very much so as skilled labor in the field. How do you see AI potentially helping with that labor shortage and also helping in the attraction of that next generation?

Subodha Kumar: I think this is where AI can play a very important and useful role, especially in the context of the US market. See US market labor shortage in construction or even trucking industry, we are seeing the real shortages. So we see a lot of white collar people filling the jobs, there are a lot of supply on that, but very few people are ready to work in the construction industry. And if you look at the data, it is very scary. We see that these days taking any construction product for individual [inaudible 00:15:50]. I was just looking at the data, sometimes you don't find anybody for six months, which is very scary and worrisome. So I think what AI is doing here is that the first of all, lot of manual labor does not need to be met. We are wasting the time of our workforce in that.

So AI, we have to identify as a repetitive task or some of the tasks, which is sorting material, tracking inventory, right? Earlier we'll say that these are the different type of art. We have very good image analytics. They can take images, they can sort things out very easily. It's out there and they're extremely good and the mistakes, it's not going to be fatal and it is much less than human. Tracking inventory, pretty good. Even paperwork, we don't need to require that much paperwork. So that's the first place I think AI is helping. The second thing is that, we have a lot of inefficiency in the construction industry. So what AI is helping now is that how we can optimize the process. So AI does not remain just a fancy buzzword. It goes beyond that. It can run nice algorithms in the background and it can solve that problem.

Another issue of labor shortage come from the bad prediction and planning or bad skill matching. We don't have a good skill matching. AI is again playing a very important role in that. It can do the predictions, it can do very good skill matching compared to what we have right now. Another reason people don't want to go to construction many times is the safety part. And I have earlier mentioned that safety can be a lot better with the AI and it's already happening.

A lot of time is wasted in training. Now with the virtual reality training processes and I have seen companies like Shell and Exxon, they are in the drilling process and all, they're using lot of virtual reality training, where you don't have to go to the ground but you get the same experience, lot safer, lot faster and much better experience. So you can have that, but eventually it is about human machine collaboration. We have to divert our workforce to the things which machines cannot do, rather than using them for the things that machines can do and also see where human can provide good assistance to machine, rather than thinking of that, they're replacing us. Actually together we can do a very good job and construction industry has a lot of potential here.

Video: I wanted to work with a supplier partner I respect.

I wanted a partner who respected me.

We needed a partner with employees who care as much as we do.

We needed a partner who values relationships.

I wanted a partner that would help me grow my business.

We needed a partner that would help us succeed. SRS Distribution delivers more than building products. SRS delivers customer success.

Heidi J. Ellsworth: One of the areas that I am, we just came back from Skills USA, a very large competition for vocational schools and we had roofing and all the construction traits were there. But one of the things that I see that is going to change, is as we have more technology on the job sites, whether that's robotics or safety or goggles or AI, we are going to need the next generation, the workforce who are coming in. Like you said, not just to be doing manual labor, but they're going to have to be actually taking care of the equipment.

And I have one example we saw recently and that was using drones to do inventory. Very much to what you said, the drones fly, they do the count of the construction site, they do everything, it goes into the thing. So who's taking care of that drone? Who is maintaining that drone? How is that going? So it's a whole new level set of how we're going to have to train people who are going to now take care of these tools on the job site that are utilizing AI or robotics. What are your thoughts on that?

Subodha Kumar: That's a very good point. See, one of the things that AI will do, like any other technical advance, more than taking away jobs, it will chase the nature of it. Rather than doing the thing, we have to deal with the machines and we can do it much higher efficiency. And we have seen this earlier when we moved from agriculture economy to industrial economy. And I will say after that, this is the biggest move we are seeing right now. Actually, we could do a lot more with much less people and the people could do this like in service industry. But it requires a lot of retooling and reskilling.

Now the World Economic Forum had an estimate that just in the US, it will several billion of expense to deal with retooling and rescaling. But in fact, the good news is that we can use AI a lot in that. AI can help us in giving personalized training to people and it can adapt very well with different level of people that we did not have. In fact, some of the projects we are working on right now. So creating more personalized education system. And here, the educational institutions, companies, as well as the government organization and nonprofit, they have to work together to see how they can use AI to provide more personal learning for retooling and rescaling. That's the one thing.

But AI can help a lot in labor training, labor retention and so on. For example, we are working with one HR company right now and the whole idea is that a lot of people are dissatisfied in their workforce because they are not recognized for their performance or they are not getting what they expect. So can we create AI for that? That's what we are working. AI can capture how you are performing in day-to-day, what you are doing well, what you are not doing well, rather than just having an end of the year recognition.

So AI can use a lot for recognition, finding out your positives and negatives so that you can be properly validated for. Also matching with the right job rather than giving you a job what you can do. All of these things, I think AI has a huge potential which will not only prepare the workforce for the future but, also keep them happier. But it comes with a challenge also, that the companies who will not be dynamic and adapting well, they will be left behind. So both the people and the companies who are not being ahead of the curve and thinking forward, they will be in trouble right now.

Heidi J. Ellsworth: I can't agree more. That is so true and I love what you're saying about actually using AI to have more instantaneous feedback on performance, on safety and all of that. And to me, that usage and those initiatives, are going to be a retention tool and also a promotion tool for employees. Again, going right back to what you just said, that the companies that are on the front end of this are going to be, they will have the best employees also, because people will want to work for them. So it's going to be a talent, it'll be drawing talent into these big companies that are progressive.

Subodha Kumar: Yeah, absolutely. And this is where I think the messagings need to go out and those who are able to capitalize on this, they will be the winners. And I know that there are some companies who are really trying to look forward, but some companies are stuck because of either their culture or the mindset. And we have seen again and again, those are the companies who do not survive the wind of this kind of changes. It's a transformative change we are going through right now.

Heidi J. Ellsworth: I was just going to say, I wanted to touch back on that, that transformative change, where we're looking at from the industrial revolution, we are now in the technology or technological revolution, which we've been in for a little while. But AI is really taking that up to the next level and is this still considered part of our technology revolution or is this kind of something even more?

Subodha Kumar: Well I will say that it is part of technology evolution, but what has really changed in last two, three years after we saw generative AI and the huge leap in image and video analytics, that is taking us to a very different level. Because still now the whole AI was built in the idea that can we give better predictions? But now, generative AI is saying that we have to go beyond that and we can generate content, not just predict the future, but I can tell you exactly this is what you should be doing, without you even asking for it. So they're going beyond human brain in some sense. So I think that is really fundamental change and no industry in the world would be untouched and the healthcare retailing, manufacturing construction, they will be on the forefront, but everybody will be touched by that. So I think it is little more than what we have seen in the last 30, 40 years because of the advent of the generative AI and some really cutting edge algorithms with the deep learning neural network and so on.

Heidi J. Ellsworth: It's fascinating, absolutely fascinating. So thank you so much. What I would love to do is to wrap up this newscast with a little look in your crystal ball, looking towards the future and where you see how business and AI will be working together and especially construction overall in the next five years.

Subodha Kumar: Yeah, I wish I had a crystal ball with me, but what I have is, we have a good research team where we keep looking at what is going to happen and I can tell you a few things that we are very confident, that we are going to see in next five years. I will not say anything beyond that, because as I told when internet came, nobody knew what will happen in 10 years. So I will not even go that route. But next five years we are very clear that there will be a clear distinction in the company. There will be some companies who will identify very well, that these are the tasks we do not need human and the tools will be available. So we'll see a lot of switching of this manual work to machine, that we will see very clearly and it's happening, but it will happen a lot more. That will lead to a lot of efficiency and productivity.

Second thing, we will see a lot of digital interface. So earlier we used to see this, we still have these billboards where you can see these ads. Now, the new method is that these ads are being sold in auction in real time. So you can say that, "Okay, next five minutes in this billboard, I want to buy these ads." Earlier traditionally, they were all pre-sold, right? But now it's a real time selling. Now that we'll see in construction industry, a lot of digital interface would come where people have to work with lot of real time. We talk about some real time example. Now because of this autonomy, what will happen is that a lot of improvement in decision making will happen. They can sort out your emails and say that you got 100 emails and this are the key ones you have to focus on.

I was talking to one Microsoft executives and they're using it in their day to day life. They said, I've stopped checking emails. It's very scary and I was teasing him that maybe that's why I didn't respond to my emails quickly, but maybe the algorithm part are not important. But it's good and bad, but at least it will improve productivity, right? It can scan files for you and so on. So we will start seeing a lot more of that.

A lot more advanced statistics, but at the same, a lot of ethical questions have not been answered yet. Will that be answered in next five years? I do not know. I also edit a journal and we are running a special issue on responsible data science. And one of the things that I was going through these papers, we have got some very good blueprint and some of them are for construction industry. Like there were some examples, people talking about predictive helmets where before falling, it can predict whether you're falling. Device at Temple University, it can predict you are falling, right? That's a great thing in construction industry.

A lot of resource optimization and virtual reality, quality enhancement. But what is missing still is that the responsible AI and ethical AI we have not resolved, and we do not know how to resolve. Who should be responsible, I'll say everybody should be responsible. We can't just say we put all the work on Google and Microsoft of the world. No, we all should be responsible and we have to see the data we are feeding is good. So I see that in next five years we will see lot of use cases coming, a lot of improvements seen, but we will also see some negative things. We will see some failures, we'll also see some bad uses, but it will be a good learning. And I will say that from that point, we will start seeing the value out of it. So this is going to be challenging lot of goods and bads in next five years in construction industry we're going to see.

Heidi J. Ellsworth: Yeah, that is so interesting. And yes, on the ethical, I mean I can just even say as a publishing group with our websites and everything, we are putting AI statements into our terms and conditions. We are very conscientious of it, talking through that. We are learning every single day. And so, for the construction industry, to think about privacy, to think about all the different things, safety, inventory, I mean everything you said, I can't thank you enough. Really great nuggets for our contractors and for all of the construction companies out there who are thinking about utilizing AI, thank you so much.

Subodha Kumar: No, thank you for having me. One last thing I'll say that please don't forget everybody in the construction industry that use AI for sustainable construction, I think that's the one thing where AI can play a very important role. So don't forget that when you're using AI more and more, I think we have a huge potential there. Thank you very much Heidi for having me.

Heidi J. Ellsworth: That sounds like another podcast, all about sustainability. That would be great. Thank you. Thank you so much and thank you for all you're doing at Temple University and for sharing your wisdom and knowledge with us here on Contractor Outlook Newscast. I truly appreciate it. It's been so wonderful to meet you and spend some time.

Subodha Kumar: Same here. Thank you.

Heidi J. Ellsworth: Thank you. And thank you all for listening. This is the type of information that we're bringing to our Contractor Outlook Newscast, something different outside the industry that we're bringing inside to really start talking about issues that will help you. Everything from politics, to weather, to AI and who knows, what is next. We have all kinds of great information coming up in the future on Contractor Outlook Newscast. And of course, you can listen to all of our newscasts sponsored by... Under our podcast sections of each of our websites. And I want to put out a very special thank you to SRS for supporting this and really bringing this kind of knowledge and information to the industries. I will look forward to seeing all of you next time on Contractor Outlook Newscast.

Outro: Thanks for tuning in to the Coffee Shop's Contractor Outlook Newscast. Learn more at thecoffeeshops.online and subscribe to get notified when the next episode launches on the first Wednesday of every month.