How To Unleash AI’s Potential in Construction Part 1

Robert Bryant: Okay. Well, hello and welcome to this special two-part InEight Insight panel discussion where we’re going to ask the question about how you unleash the potential of AI in the construction industry. So today we’re going to be setting the scene and asking three very real people, not a machine, how do we prioritize and prepare for AI in construction? So before I introduce our panel, I wanted to share with you some quick housekeeping to make sure you get the most out of the session today. So first of all, as you come in, you are all automatically muted just to help us keep the channel and the audio nice and clear. So this is a one-way broadcast. However, there are ways for you to interact with us as we go through this webinar and certainly to pose some questions. So if you are having any issues, please use the chat and message us and we can help you through any troubleshooting. But more importantly, we are very keen for you to ask questions throughout the webinar using the Q&A feature. So you will see that on your screen down in the bottom corner. So please do enter your questions as we go through, and we’ll do our best to accommodate those. We’ve got a lot that we want to talk about, but please do put them through, and we’ll do everything we can to make sure we address those for you. Just a very quick message, for those of you that are not familiar with InEight, we are here to challenge the industry on just how much you can improve your project outcomes through the use of data and making data-driven decisions. We do that through the entire project life cycle. So there’s a number of aspects of that all the way through from the beginning of the project where you are in that planning stage all the way through to the handover point. So if there’s anything that you’d like to find out about InEight, please do visit ineight.com to find out more. But we’re here today to talk to our panel and to get some insights from them on all of the things that we want to learn about AI and exactly what AI is all about when it comes to the construction industry. So it’s my great pleasure to introduce our panel. We have with us today KP Reddy, who is the founder of Shadow Ventures. We have Salla Eckhardt, who is vice president of the innovation for OAC and Sam Zolfagharian, who has a PhD and is a chief technical officer for YegaTech. So we have three people here who are very well-equipped to talk about this topic, and we’ve had some great discussions in the lead up to today. So really looking forward to sharing with you everything that we can in the course of the next hour. So I’m going to ask our panel to introduce themselves briefly before we get started and just to give you a bit of background and context on what they’re going to be able to contribute to today. So perhaps, KP, you’d like to start and we will go around the virtual screen here. So KP?   KP Reddy: Okay. Thanks for having me. So KP Reddy, founder of Shadow Ventures. I’m a second-generation civil engineer, started writing code when I was 13 for my dad ’cause he made me. Practiced as an engineer and then went into startup world building a couple of companies when I was fortunate enough to take public, so that gave me some optionality, did a lot of work in the BIM space and then about five years ago decided I wasn’t up to the grind of building another company, so decided to become an investor instead. Shadow Ventures is back by almost 100 strategic investors that are all from the construction engineering and building product space. So our idea is that we don’t know everything but our investors sure do.   Robert Bryant: Fantastic. Yeah, I love the perspective that you bring and looking forward to it. Thanks, KP. Salla, if you’d like to tell us a little bit about yourself.   Salla Eckhardt: Thank you, Rob, and thank you InEight for inviting me for this discussion. It’s great to be here with KP and Sam. I’m Salla Eckhardt. I’m the current vice president of innovation at OAC Services. We’ve been around for 68 years being owners rep for matriarch real estate owners and developers and helping them succeed with their projects. We do construction management and program management in the United States. Pleasure to be here. Those that don’t know my background, I’m an architect by training, but it was very early on that I pivoted my career to become a technologist and an industry pioneer for emerging technology. So AI is now one of the key topics that I’m focusing on seeing which direction we are taking it with the build environment industry. So thank you so much for having me here.   Robert Bryant: You’re welcome. Thank you for joining us. So I know from our discussions that you’ve been challenging the use of tech yourself throughout your career. So again, looking forward to the perspective that you’re going to bring to our discussion. Sam, welcome to you.   Sam Zolfagharian: Yeah, thank you so much for having me here as well. I’m excited for this upcoming conversation about the potential AI. My background is in structural engineering. I used to design and build residential buildings, but later I decided to go back to school. Back to my time at Georgia Tech, I was studying construction management and construction technology. I was involved in national BME standards, see how we can manage the data exchange models between different stakeholders who are involved within a project. Through that project I got an offer from Autodesk. I went back there and at Autodesk we were trying to see how we can have a similar approach as BIM, it’s called MIM or manufacturing information modeling for the manufacturing industry. So that’s what was my main focus back at Autodesk. Also, later I was working on industrialized construction, how we can bring rule-based automation system into the design and modular fabrication. Later, we included AI as well in terms of tailoring our users’ experience based on who they are, what they do. Right now, I am the CTO at YegaTech. We’re helping the design and construction firms with their data and AI strategy while navigating its risk.   Robert Bryant: Fantastic.   Sam Zolfagharian: Thank you so much.   Robert Bryant: Thank you very much.   Sam Zolfagharian: Yeah.   Robert Bryant: No, that’s great. That’s great. Look, I think what you’ve all done is provide some superb context for our audience to understand exactly where you are coming from and how well-equipped you are. So we’re looking forward to getting into this discussion. I should mention too for our audience that we do have some polls coming up throughout. So as we get into this, you’ll see some of those come up on your screen. Please do take part in those. It helps us gauge who we have on and what your level of interest is and where your thoughts are as well. And we can play that into the discussion that we have, so very interactive chance for you to take part in those as we move through. All right, let’s get started and get into our discussion. The construction sector is recognized generally for being, if I politely put it, a little bit slow at adopting technology as a vehicle for change. So I think most people will say there’s a lot of technology in the industry, but I think as a vehicle for change the industry has been quite slow. It’s been reported as one of the lowest industries when it comes to investing in technology way down the list in many years behind forestry and fishing in terms of its investment and utilization of technology. It’s been highlighted as one of the least productive industries. When you look at the industry, gross product for hours worked. So with those things in mind, you can take them either way as positives or negatives when it comes to opportunity, how ready is the industry for AI? Are we seeing early adopters? Have you seen the impact coming through already? So just a couple of small questions to kick us off and warm up the discussion. So I’ll start by asking KP for your perspective on that, and what is your take on AI today, and where do you see it going?   KP Reddy: Yeah, I think what’s important to understand is the leadership of these organizations. They may not be digital natives, but they all started work with a computer in front of them, so it wasn’t like a big drastic change in terms of the world of compute. Our hypothesis is generally that not since the Blackberry has the CEO of a construction firm cared more about a technology, i.e., they have it in their hand, they have it on their iPhone, they’re playing with it. They might be spending more hours in ChatGPT than the CIO is, and so they have questions. I think as they look at the vision and the future, it’s being the most experienced people in industry, they actually have a very strong vision for where things are going and are now questioning do they have the team that can actually execute on it? Do they have the vendors that can actually execute on it? Which I think is like, it’s a magical time from that perspective. There is no convincing, I don’t know any CEOs that were pulling out a copy of Remit and playing in BIM and going, “Oh, my God, this is amazing.” They were like, “Hey, that’s the CAD guy’s job.” They don’t care. This is the first time that they actually care. My phone rings off the hook saying, “KP, this AI thing is amazing. I can see it. I can see where this industry could go,” and I am super excited about it. I think to your point about we’re behind means that there’s also an opportunity, the incremental change in the opportunity for me as a VC, selfishly, the opportunity for incremental change is massive and therefore, it’s a massive financial opportunity from my lens. So I think that that’s really happening, and they’re just questioning, “How am I going to get this done? Are my software vendors going to give me a Copilot plugin and now all of a sudden I’m doing AI or is my team supposed to build? I’m not sure, but I know this matters.”   Robert Bryant: Yep. Yes, Yes. That is encouraging. I think that’s a nice, good positive perspective take on it and opening up the value. Sam, what’s your take?   Sam Zolfagharian: I want to highlight back on KP’s point. The industry is not reluctant to technology. They have their phones, they try to adopt the solutions based on the roles that they have. I also think at this time, the difference is people think that AI is the hype. They’re asking about AI and they’re saying, “Is it the hype or not?” It’s not the hype, it’s just because of the opportunity that we have right now, we have higher computation power. The chip technologies are impressive. The amount of operations that they can run in one second is massive. It’s like having a supercomputer with 1.1 quintillion operations per second is out of our capabilities and more computation means more data. Also, that enables better algorithm because we have more data to train our AI algorithm. So the combination of this is enabling a brighter future, and I see ChatGPT or large language model as a disruption right now, but I believe that there will be another disruption coming up in the next two, three years. So it will be like an ongoing process because of the power of computation data and algorithm that we have. As a result of that, the AI will be democratized. It means it’ll be accessible for more people and it will be cheaper. The technology building development will be cheaper and more accessible, and that’s where a lot of AI solution will come down the pipeline. We should watch out even in our industry how we adopt those technologies, how we react to those technologies and how we can make sure that we have policy regulations in terms of adopting trustworthy AI within our company so that we can get the highest benefit. So my point here is AI is there, it will be adopted within our workflow, and it’s not like a hype right now.   Robert Bryant: That’s something I’d like us to get into is obviously these priorities, so there’s a lot of opportunity to take. So we’ll explore that a bit further, Sam, because I think there’s a lot of good things in what you’ve just said there. Salla, your take on this. Where are we today?   Salla Eckhardt: Yeah, I think we’re early on as an industry to adapt to AI. When thinking about the industry as a whole, the reason why people chose the build environment industry and construction industry is that they wanted to work in person with other people and work with their hands. So it was pretty much to summarize the people that didn’t want to have anything to do with computers to begin with that chose this career track. It was for people that could get a new job or a first job in construction without any education, and that’s the history of our industry. There’s a lot of tribal knowledge within the industry ’cause the longer you work in construction industry, the more you know, the more are exposed to different types of situations and solve things. Construction is always varying up to weight. So when thinking about emerging technologies and it’s adoption rate, it always comes down to the question that, is the technology faster than the human being? If it’s not, then it’s not ready. When thinking about AI and how it’s being used for solutions as narrow AI, that you can get an instant answer for any question that you have if you know how to formulate the question so it makes things faster. You don’t have to pick up the phone and wait for an answer from someone or get the, “I don’t know,” answer. So people feel more supported with the AI platforms ’cause they can get the instant answer for something and continue from there. But we are very early on, we don’t have enough proof of where it could be applied in a way that it’s trustworthy. So we still need to figure out the rules of the game and make sure that the things that we are doing are based on global standards. That’s something that I want people to understand that the standards are basically something that we need to address as a law and then create the playbooks and the guidelines that are more organization based to get everyone play towards the same goals. But the standardization work needs to be AI ready.   Robert Bryant: Yes. Yes. Yeah, and that resonates well with me, Salla. I think the idea of standardization and having structure in the data and that data literacy is so critical as we look at this. So what I’m getting from you all is that we’re really in a positive position. There’s a lot of opportunity to be taken. There’s more enablement available that the technology can bring genuine improvement perhaps more than it has in the past where it may have been perhaps just a bit of a fad, that we’re actually in a point of some real benefits to be realized. Also, that there’s a greater awareness through the industry all the way up to the executive and decision maker level around how technology can and should be used. So all these things suggest that we are poised and really at the beginning of a lot of change for the industry. So perhaps it’s a good time to ask one of our first poll questions as well. So for the audience question for you, to what extent do you believe AI can improve project productivity? I think this is a critical one for the industry, and we talk about the topic of productivity, and our panel have touched on that just now in terms of where it can go. So whilst you think about your answers there, and we’ll bring the results up shortly, but give you some time to answer that question. I’ll put the question across to our panel around this, what are the productivity gains that you think we’ll see? So based on your take of where the industry is, the interest that’s there from the executive level, the untapped potential to realize the value of business and all of the gains that are poised for us to take, where do you see productivity gains coming? So I’ll have to pick up with you off the back of your comments before. So where do you see productivity gains, Salla, as you look at this for the industry?   Sam Zolfagharian: Are you asking me or Salla?   Robert Bryant: Salla, can you give me your take? Salla. Sorry, Salla.   Sam Zolfagharian: Oh, yeah. Yeah.   Salla Eckhardt: Sorry, I missed that.   Robert Bryant: That’s all right. That’s okay.   Salla Eckhardt: I think a lot of the productivity gains will come from all the checking and verification that construction industry people have to do. There are so many different players with a different perspective into how to solve and what to do next that we are constantly verifying and checking the data that is given to us. We are constantly also thinking about is it trustworthy? Is it the digital truth that we are looking at? So overall, I think that a lot of the productivity gains can come from the simple busy work going away or the busy work being taken care of with the computer algorithm and AI and people concentrating more on talking about what the next steps are and what are the common goals, and what is the core process that we are aiming to follow. That way, it is more proactive approach into the projects management and project control rather than always putting out the fires that we’re used to doing.   Robert Bryant: Yes. Okay. Yeah, so that’s nice. I think the proactive approach. Sam, sorry, I was a bit unclear in my pronunciation, but what are your take on it in terms of productivity gains?   Sam Zolfagharian: So one way that I will look at how AI can help us is make us more proactive rather than reactive. Because again, back to that point, the computation power, we can have a holistic view over the whole project and see what’s going on because there are tons of point of data throughout the project that we need to analyze, and it’s out of our human power, and that’s where computer can help. People are saying, “Oh, digital twin was a hype last year, and now it’s AI.” Digital twin wasn’t a hype and digital twin is there, and we were using digital twin like something like open space, how they can help the construction companies to compare the reality versus as-built versus as a plan and making sure everything is monitored. Now we can take it to the next level with the power of AI. We can include large language model, we can bring that data from our documents into that data analysis so that we can be more proactive rather than looking at some numbers. That’s where AI can help us with combination with digital twins. I want to get back to Salla’s point earlier about people would like to be on the construction side. They learn, they talk to others, and that’s a positive side of being human. We like connection with other humans. On the other hand, there are some data that is accessible or more analyzable, I don’t know if it’s a word-   Robert Bryant: Yes.   Sam Zolfagharian: … by computers.   Robert Bryant: Yep. That’s okay.   Sam Zolfagharian: So the combination of partnership between human, what we know, what we learned throughout the project, talking to others and what computers have visibility on can help us to have a positive productive partnership and learn from each other. So it’s not just computers, AI algorithm learning from us to take over the whole, this is what people think is also us growing with the feedback, with the insight we’re getting from AI algorithm and also in talking to other people too. So I’m looking at it as like a partnership productivity approach.   Robert Bryant: Interesting, Sam. I think we’re here starting to get into a question that Tariq has posed in the chat, which was the areas that AI will be applied to. I think if we’re looking at it here, we’re talking about how AI can help open up all of that human computation and what we might call the industry knowledge or the organizational knowledge that can be tapped into through AI to make it available to more people within the business and even within the industry, but certainly within the business, so that they have that at their fingertips where they might not otherwise have that. If those people aren’t sitting shoulder-to-shoulder with them, that organizational knowledge can still be captured if your data is there. If the data inputs and the patterns of behavior and the thoughts and decisions are captured, then AI can help to identify those patterns. I think that’s what you’re talking about, right, Sam?   Sam Zolfagharian: Yes, and you are getting into information retrieval. Like when we’re talking to customers, one challenge is when they hire out of school, like junior engineers, they have to work with a senior engineer to see how they design, how they operate on a project. How that knowledge from those people who might be close to be retiring or even some of them now is really common. People leave their job looking for other opportunities, how we can capture that knowledge for junior engineers so that they can leverage it, they can have that internal information retrieval, something like ChatGPT, but for my own company, for that-   Robert Bryant: Yes.   Sam Zolfagharian: … secret sauce or IP that I have internally.   Robert Bryant: Yeah, fantastic. There’s another question that’s come up in the chat. I think, KP, you’re probably poised to answer this too when it comes to examples because I know that in your exploration of opportunities for investment, you are seeing this in businesses. How are you seeing productivity gains across the project lifecycle, including that onsite construction as much as perhaps the office side of it? So applications of AI to improve productivity in the construction at all points, including onsite. What’s your take, KP?   KP Reddy: Yeah, so I think it’s important. OpenAI is a 15-year-old overnight success. I think it’s important to understand that. ChatGPT is to what Netscape was to the internet. In other words, from accessibility. AI has been, it’s not a new thing. It’s more that now it is approachable, accessible at a cost that is attractive to construction ’cause we can’t overspend on things, we’re cheap. So I think it’s important to understand that context. Also, AI that I’ve built in the past and investments we’ve made over the years, it’s actually not about structured data, it’s about the ability to infer and make inferences off of lots of unstructured data. So I think that’s the challenge that a lot of in construction, it’s like, “Oh, we have all this data,” where people run around, “Oh, I have the data, I have a data lake.” It’s like, “Okay, that means nothing.” It actually has nothing to do with that. We all have lots of data and the heavy lift is how do we make that data usable, which has been very difficult because it is highly unstructured. So the whole point of AI is absolutely about unstructured data, not about structured data, and that’s the benefit. I think the accessibility to that is important. I think also limiting it to knowledge bases is not a great… that’s a small part of it, right? The knowledge bases-   Robert Bryant: Yeah.   KP Reddy: … of LLMs and all are a tiny, tiny part of the opportunity because you have to have to look at equipment on site and what’s happening there. So I think the productivity is multiple levels. For construction, I think we’re quick on productivity. I don’t think anyone’s walking around with a stopwatch saying, “How fast can that guy handle a nail gun?” Right? I don’t think-   Robert Bryant: Sure.   KP Reddy: … that’s the slicing we’re trying to do ’cause I’ll tell you what-   Robert Bryant: That sounds dangerous. Yeah.   KP Reddy: … I’ve seen people that can operate a nail gun faster than any robot, right? It’s amazing.   Robert Bryant: Right.   KP Reddy: They’re talented people, right? But the difference is when we look at robotics on the jobsite is using AI and computer vision to have lights out, full autonomy, i.e., I’m done for the day. I hit the robots on and they’re going to work while I went home. So that full autonomy and contextual understanding of the jobsite and what needs to be done, that’s a real powerful part of AI in terms of productivity. I’ve built robotics companies, I’ve never seen them faster than the best human, right?   Robert Bryant: Right.   KP Reddy: But they will work 24/7 in a lights out environment. Oh, by the way, we had this thing called safety that we all are very… it’s very important to our industry. There might be some jobs that we just need to stop letting humans do ’cause it’s just not safe, right? So let the machines-   Robert Bryant: Right.   KP Reddy: … go do that work. So I think the productivity gains are massive. On the knowledge sharing, if you take LLMs as the base layer to manage and interpret your unstructured data, then you have to layer on expert network data that then begins to inform for specific situations that are informing and teaching the system. So I think a lot of CEOs have asked me, “Oh, so I take all my data and throw AI on top of it, and now I know everything.”   Robert Bryant: Right. Right.   KP Reddy: It’s like, “Not exactly. Not exactly.” So I do the idea of how we think about productivity is expanding our point of view of productivity. My last point is, the number one thing we all get caught in on construction jobs is the interdependency of tasks. We work in silos because we don’t like sharing the truth quite honestly. The drywaller doesn’t want to say, “Hey, I’m behind.” Like, “Hey, we’re fine. We’re on target, we’re getting there.” Right?   Robert Bryant: Uh-huh. Yeah.   KP Reddy: Nobody really wants to share. But I think with AI tools, I say it’s like my kids when they’re out, I’m like, “Share your location.”   Robert Bryant: Yes.   KP Reddy: It’s not that I don’t trust them, I just want to know, right?   Robert Bryant: Yes, absolutely.   KP Reddy: So AI has the ability on the jobsite to be that share location tool so maybe we reduce the stress of our subs that they don’t have to deliver bad news, that the system’s already getting the bad news and interpreting it and giving them… ’cause what we care about is what’s your next best move, right?   Robert Bryant: Right.   KP Reddy: We don’t care that you’re behind, it’s like, but what’s the resolve to you being behind? How do we get back on track? I think I really see a lot of productivity gains there. Come on, the whole RFI system is just the dumbest thing ever, right?   Robert Bryant: Right.   KP Reddy: Get rid of that, and you have so much productivity gains.   Robert Bryant: Yes, yes, yes. Yeah, the time taken to fill in the request for information to tell you that this is the current state when you could see the current state more readily with the use of AI.   KP Reddy: And you probably may have lied in the RFI too.   Robert Bryant: Right. Yeah. Yeah, being optimistic. So, okay, that’s a good point. That’s a good point, KP. I think we’ve got some good questions coming through that are getting addressed as we explore this conversation as well. But we will make some time towards the end to make sure we capture them. So please do keep them coming through. Really enjoy seeing them all pop up. First poll, what do we have in terms of results? What are people seeing and saying? Okay, so if I can scroll through that. So if I see it right, the majority quite clearly are saying there moderately — more research and testing needs to be done. Okay. All right. So a understandably skeptical audience in terms of that question and what people believe AI can do to improve productivity on projects. The majority of our audience is saying that there’s a moderate impact and more research and testing needs to be done. So we might visit that as we get through in our next question and explore why do we think that more research and testing might need to be done? Is it anything to do with the data quality? Are people perhaps a little bit cautious about this knowing that there is work to do in order to generate good outputs from AI? We’ve talked a lot about the importance of data health and I know, KP, you were talking previously about the way AI can address unstructured data. I think we might want to talk about what that means, the role that it plays in operationalizing artificial intelligence if we talk about data health, how you leverage it, how important it is. So if we can explore that question, and there’s a chance for us to also pose another question to our audience. So as our panel is getting ready to address my question about the importance of data quality and connected systems, a second question for you. So we’re going to make you work hard as an audience. How good is your company at collecting and analyzing data? So you can think about that and answer that question as I turn to our panel and ask them to answer the question, what are your thoughts about the importance of data health and the role that that plays in being able to operationalize AI? ‘Cause sometimes AI gets labeled as this silver bullet of, “Hey, just…” I think you said it, KP, “just throw AI at my data lake and boom, magic comes out.” So what’s the reality of that? Salla, I know you’ve had a lot of experience through your career in pioneering technology, so I’d like to start with you on this question. What do you see as the answer here around data quality and connected systems?   Salla Eckhardt: Thank you. The data quality really depends on who is the creator or the author of the data. It comes down to the question that, do you trust the data that I’m producing for you to use, or do I trust what you’re producing for me or for others? If there is the habit of putting in placeholders or just putting in something that is a strong opinion or even a vague opinion, just a gut feeling that this might be close enough, that’s where things already start to steer off and it’s really hard to go through all the data sets and figure out which ones are the placeholders that are the copy-paste of past projects or someone who is the know-it-all that everyone goes to for answers. Since they are put in that position, they are very eager to share their opinion on things. So it really comes down to people’s behavior in terms of data authorship and creation, and then we have the data owners that have their stake in the game that might be a mandating certain data or a way of producing the data that serves their purposes. So there’s all kinds of ethical questions regarding the data management. If we are dealing with a lot of data that is years or decades of accumulation, then I don’t think that a lot of companies had very sophisticated data master plans or data strategies way back when. I don’t know if they still have really sophisticated strategies for that. So then my uncertainty in terms of analyzing data or doing anything with the data really depends on the trust in the industry, that do we trust the data that others are creating and are we ourselves trustworthy for the others? Because like KP said, nobody wants to reveal what they did wrong and people will try to do other things.   Robert Bryant: Yeah. Okay. So building trust in the answer. What are your thoughts on that, Sam? How do we go about that, the question of data quality and also this issue of trusting what you see in front of you?   Sam Zolfagharian: So one way we can think about it is for the last at least 20 years, we’re using multiple solutions out there, but we were concerned about our data. When we go to an Amazon e-commerce solution or using Uber app ride-sharing apps for getting to our destination, all of these products, they have our data. They know exactly who we are, our demographic. What’s happening with AI, though, we were concerned about our data privacy at that time too. But right now with AI, it’s getting more into the data ownership concern, how we can manage the data ownership concern. We all know in a construction project, multiple stakeholders are involved, and it’s difficult to figure out even in a BIM model who owns which pieces of the model, who owns the data? When we are going to leverage that model for building an AI system, that’s getting tricky. Do I have the permission from my client or not? Or does this part belong to that mechanical engineering subcontractor or not? So one risk here is the data ownership, and we are hearing it a lot from the news as well. The difference here between AI and those past solutions websites or these other apps that we’re using is the data longevity. So what it means is right now based on GDPR or other regulations, I can ask company to delete my data whenever I want. At least most of them are committed to do so. But now with AI, they can generate insight from my data, they can build a new solution. Even though they delete my data, but there are some insights there that are extracted from my data has been used for that training. Like company tech providers, I don’t want to mention a specific one, but tech providers, even if I ask them to delete my data, what would happen to that insight? Who owns that insight generated? So that’s where the trust will get a little bit … Again, back to Salla’s point, in terms of human bias, we are biased. We have about 168 bias on average in ourselves through their conscious or unconscious. We’re injecting those when we are generating data into our work, whatever we do. Also, even building the AI algorithm like those engineers, AI engineers or tech providers who build a solution, they may use a specific database to train that AI model, and it could have some bias in it. So we need to watch out for those risks as well that can come from either human data or algorithm themselves.   Robert Bryant: Yeah, that’s an interesting one because I wonder, could we say that AI helps to keep us honest as well on a project? I think there’s a nice thing that KP opened up there before about the reality of a situation. If the technology is helping us to track the actual situation and what’s truly being delivered and completed, can AI remove some of that bias as an obstacle in achieving realistic project planning and scheduling to say, “Well, here’s what you think your project timeline should be, but in reality, here’s what your project timelines have been according to schedules for delivery,” and allowing people to make guided decisions. So helping to build trust in that way to say, “Look, here’s the transparency of what you think you are going to do versus what you’ve historically done as an organization or on this project,” and allowing people to make an informed decision. Is that perhaps some of the answer KP to how we can build trust in AI as well as improve and leverage off good data quality?   KP Reddy: So there’s a couple of things. One, I think as we think about this, we can get into these binary ideas. So I look at email, there’s some people that put all their email in tight little folders and their desktop is nice and organized. That’s not me. I just throw everything in the bucket and search, right?   Robert Bryant: Yeah.   KP Reddy: So there is some benefit of the ability like search across unstructured data and less of a relay in view of the world. You can just ask it questions and it’ll go find that email kind of thing, right?   Robert Bryant: Right.   KP Reddy: So I think it’s not this binary thinking around that. I think Sam brought up a good point. I’ve been advising a lot of construction firms on data privacy and policy because no one is clear. Plans and specs are copyrighted by the designers for your best… You’re giving the license to use it to perform a function, i.e., build a building. They did not license it to you, or maybe they did, it’s submitted to inform your AI engine. It’s their data.   Robert Bryant: Research.   KP Reddy: So I’ve heard some of the companies that Sam was [inaudible 00:40:30] I’ve talked to all those CEOs. What they’re saying is, “We’re not going to do any of that right now.” The main reason is if you’re going to, if I’m the architect, I’m going to ask for something. I’m going to say, “Hey, I will give you the right to use my data to inform your AI model. However, I need a penny every time it’s… for every bit,” whatever it is. So I think much like the saying that if you’re not paying for something, then you are the product, right?   Robert Bryant: Yeah.   KP Reddy: I think you’re starting to see that theory. But the problem is the big technology companies aren’t sure what they have to offer, so they’re opting out right now. We’re being very explicit with the CEOs I talked to that are saying, “We are not using your data to teach any AI models or anything.” But I think as soon as they understand what that business model is to tell you, “Hey, can I buy all your drawings to inform my AI model?” Then that might change. You have the optionality to do so, right?   Robert Bryant: Yep.   KP Reddy: So everything’s for sale, everything’s for sale. We just don’t know how to charge for it. We have to figure out how to charge for it. So I think that’s what you’re seeing on the data side there about monetization and right to use these data models. I think that’s actually happening pretty fast because I think probably anybody that’s running a Copilot plugin is excited and ready to start sucking in more data to inform their model. So I think we’ll get there soon-   Robert Bryant: Yes. Yes.   KP Reddy: … but I think the software providers are going to come… have to have something to offer in exchange.   Robert Bryant: Yeah. Yeah, that makes sense. That makes sense. I think it’s back to one of those early points you made, I think about realizing the value of an organization in different ways. So it’s being able to tap into their intellectual property that is the way they operate, right?   KP Reddy: Right. But Rob, one last point, the person we haven’t-   Robert Bryant: Yeah, go ahead.   KP Reddy: … been talking about is the actual customer, right?   Robert Bryant: Right.   KP Reddy: The owner, the developer, right?   Robert Bryant: The owner of the asset, yeah.   KP Reddy: So the asset owner, I’m having a ton of conversations with them and they’re saying, “How do I use AI to make these people accountable?” There’s not a deadline that they haven’t-   Robert Bryant: Right.   KP Reddy: … missed. There’s not a budget that they don’t blow right through. They’re terrible to manage as an asset owner. I want an asset built that generates cash for me. That’s what I want.   Robert Bryant: Right. Yes.   KP Reddy: All these other architects, engineers, contractors, they’re just in the way. They’re in the way of me getting my job done. So how do I use AI to hold all those people accountable? ‘Cause I don’t know anything. Right? So I’ve had-   Robert Bryant: Right.   KP Reddy: What I want to start doing is running all their drawings and schedules and everything through an AI engine, and I need to know better than they do what the bad news is.   Robert Bryant: Ha. Yeah.   KP Reddy: I want to have accountability. I want to have governance over this black box called design and construction that I spend billions of dollars against, and I literally have no idea what I’m getting. I don’t know what I’m getting-   Robert Bryant: Well, you’re right-   KP Reddy: … and I don’t know what it’s going to cost.   Robert Bryant: We’re used to the industry operating, as you say, in a black box with missed deadlines and blown budgets. So perhaps without being flippant about it, there’s actually more trust in the technology than there might be in the industry today anyway if you can start to apply some analysis to the data rather than just relying on the report. So yeah, okay. Okay, nice. So maybe we can trust the robots a little bit better.   KP Reddy: Absolutely.   Robert Bryant: How did our audience go? So our second question to the audience, how good is your company at collecting and analyzing data? Looking through, again, very clear answer. The majority are saying fair. 68% say that they have some structure, they manually export from multiple systems, and they use it for some reporting. So again, I’d say that’s encouraging. There’s a small percentage, let’s say 12, who are very good at it with consistent structure, but 68% believe that their organization is fair, maybe skewed to the profile of people that are logging in and taking part in this poll anyway, but good to read the answer and good to see how people are approaching AI. All right. We have a little bit of time left, and I do want to make sure we maximize that with our panel and get some more insights from them. So I think one of the most useful things that we can do for our audience is to give them some further insight on what steps their organization should take. So this is a burgeoning area of the industry and being able to maximize and leverage the application of AI, and after all we’re talking about today, prioritizing preparation. So if you are looking to set up your organization and take it from, as our audience have told us, fair to very good, what are the steps that they can take in their organizations to become AI ready? Whether you’re just starting to develop a strategy around this to explore how you can use solutions or what are some of the steps that they can take? So KP, come back around and start with you on this ’cause I’m interested to hear what you are seeing in perhaps some of the best businesses that you are engaging with and what you’re seeing overall. What are organizations doing today to set themselves up to leverage AI?   KP Reddy: Yeah, it’s interesting, and I think I shared a link for our Getting Started With AI white paper. But I published that and because I look back in the internet days, I was getting the same question, “Why does Steve in accounting need a web browser? He only needs email. What’s he going to do with the web browser? He’s going to goof off. He’s going to waste his time. He might create a security breach.” So all the narratives I heard with the advent of the internet back in ’96, ’97, I’m hearing the same things about AI. Being the old guy, my advice is the same. Train your people to be power users on the available tools that are out there.   Robert Bryant: Right.   KP Reddy: Create pods of people that are the power users that are training other people. Funny enough, we used to have to train people on how to use a web browser. That seems unbelievable today, right?   Robert Bryant: Mm-hmm.   KP Reddy: But it was like, “Here’s how you connect to a modem.” It was that type of thing.   Robert Bryant: Yes.   KP Reddy: I think the best thing every organization of any size can do, whether it’s, “Hey, we set up a committee,” whatever your method is for deploying knowledge to your teams is training them on the tools that exist and building the vocabulary with them and helping them understand what is Midjourney? What is ChatGPT? How’s that different than this thing and that thing? Just get them to be power users so they start building comfort and then they also on a one-on-one basis start to experience the gaps. They start to experience, “Oh, that’s weird.” I like to pick on marketing departments ’cause marketing departments and construction companies probably the least amount of risk versus estimating. But if the marketing people, the next proposal they have to put together starts using an AI tool to build the proposal, they’ll be like, “Oh, that’s weird, that’s wrong.” They start to experience what’s right and wrong and the limitations and the opportunities are at a one-on-one basis and really do that. I think that is the best thing that every organization, big, small… Our firm, we have about 20 something billion of GC volume that work with us. A lot of those are large firms, and they’re doing these things. But I think even the five-person, 10-person firm, not knowing how to use AI tools is like not knowing how to use a web browser. Shame on you.   Robert Bryant: Yeah. Yeah. Yeah. Interesting. Yeah, we’re certainly at that point. So catch up fast, learn how to use the solution, learn how to apply it to your task within the business. Salla, what’s your thoughts on this? What are you seeing? Again, I always remember the story you told about your early adoption of technology when you were working in architecture, but what are you seeing as something organizations should be doing now to be AI ready?   Salla Eckhardt: Yeah, thank you. So I’ll take the industry 5.0 perspective into this that the foundational first step would be to really figure out the data on cybersecurity so that the people don’t go off rogue and just do their thing and then be like, “Oopsie, I didn’t see that happening,” but really having the data security strategy and have that educate to people that please don’t do something before you understand what you’re doing. But then the second step would be to take that industry 5.0 approach that AI enables the tools to be applied and customized for the different context of different end users so that we don’t have to teach people, “This is how I use AI, please do it the same way,” and that way tie their hands and have them wonder, “How is this related to my role in this company or this project?” But really addressing it as a platform thinking that we might all have the same tool, but then use cases depend on what your needs are and really go in and talk to people like, “What are the inefficiencies in your everyday work that we can then solve by giving you this tool into your toolbox?” And really opening up the discussion that, “Where are the inefficiencies? Where are different organizations burning their budgets and money, and how would they benefit from the new technology?” Rather than tell them how they should be using it and then facing the pushback that, “This doesn’t really apply for me, but thanks.” So then the third approach or step is that once you have clarity on an uncertainty that people are not using the tools under the table hidden ’cause they want to use it, but they don’t have the data security readiness or the organization hasn’t given the guardrails for it. Organizations need the data master plan ’cause now, even in less than a year, AI-based solutions have popped up like mushrooms after a rain that there are so many different applications that nobody can know how to do and how to use all of them. But with the data master plan, there’s a long-term strategy into what is the core process that we are trying to resolve? What are the core functions that we want to support? Then look for the right technology and the tools that then help create solutions rather than solve one issue and create five problems.   Robert Bryant: Yes.   Salla Eckhardt: So data master plan is something that I highly, highly recommend for all organizations that are looking into emerging technologies.   Robert Bryant: Yeah, and that’s a great point. I think it’s one of the things that I think it’s really important that it’s considered at a strategic level within business so that they consider it in the same way they do other aspects of their organization and make sure there’s a consistent approach. That way, people know it’s being considered. The early adopters can be part of that journey and get into it rather than doing it, as you say, under the table and then others understand why they’re being asked to apply that technology to the decisions and the processes. So I think that is a very important one. Thank you. Sam, I’m going to come to you in just a moment, but I do want to ask our audience for their inputs one more time. We have our third poll question, which is, what are your greatest concerns about AI in construction? So if you are skeptical or if you are concerned, if you’ve been following this for some time and you feel that there’s something that people need to be aware of, there might be an answer in this. What are your greatest concerns about AI? We’ll come back to that in just a moment. We’ll give you all a chance to answer that question. But Sam, back to you on this question of organizational readiness for AI.   Sam Zolfagharian: I would share a little bit about how we work with architecture, engineering and construction firms right now. So AI we all know is just a tool. When we are working with one of the architecture firms, it’s the largest one, SSOE. The [inaudible 00:54:10] says that AI is not our strategy, AI is the accelerator for our strategy. To achieve that, we are thinking about it in terms of people, tech and process, because tech is not the whole problem. People are involved as well. So in terms of the tech and data, people usually go, companies go and build like a data lake. But in terms of AI, we need to think about where are those AI opportunities for our companies and which data we need to achieve that. Otherwise, data is a rabbit hole. We can just capture, capture without knowing how we are going to use it. There is an example, like an analogy in the augmented book that is talking, it’s like preparing dinner for your guests. You should know what you want to cook and then go get the ingredients. So it’s similar here. We need to know what AI opportunities we are looking for so that go and get those data and think about policy limitation, the data assessment part of it. In terms of the people itself, back to Salla’s point, it’s like AI solution may come over and they’re like, “Oh, I don’t know how to use it.” So education, continuous education really, really matters, because AI, people should know they shouldn’t over rely on AI solution. They should know that they should partner with them. If you just over rely on whatever insights we are getting from AI, there could be a bias like missing transparency or a black box situation. But if we have that partnership and people are educated in terms of how to partner with AI, that’s where you can get value, and people can grow in terms of with the rate of advancement in AI itself. That’s where we are helping the companies with AI education as well. Last but not least is the process too, because now we can remove the bottlenecks, the repetitive task from our process by educating people, by leveraging some AI solutions, how the new workflow may look like. It’s more than just automating some tasks. It’s even about finding opportunities, something that wasn’t possible or even we were not thinking about it. Now we can include it in our proposal so that when we can submit it to our client, we can shine in front of them. So we need to think about AI not just in terms of automation, but also augmenting our business model.   Robert Bryant: Fantastic. Thank you, Sam. Somehow, time has accelerated, and we now just have a couple of minutes. So I want to cover off a couple of things. First of all, I do want to see what our audience thought, so we’ll quickly switch to that, are your greatest concerns about AI in construction, a little bit more broken across the responses. Looking across here, most people are saying that it’s the lack of transparency and a loss of humans in the decision-making that they are concerned about just ahead of privacy and security concerns. At the top it’s split, so time, resources, and expense to implement. So it sounds like people are concerned about the challenge of implementing it and perhaps people being avoided in the decision-making. So I want just if I can get 30 seconds from each of you on your response to those responses, those answers. What are your thoughts? How do you make people feel a little bit better about what they’re frightful of? KP, 30 seconds from you on that.   KP Reddy: Yeah, I think it’s about getting better educated about what it is. I think this is all iterative. It’s not decide overnight, deploy a bunch of resources against it overnight.   Robert Bryant: Yeah.   KP Reddy: I think we are such a risk averse industry, we know how to put in governance and checks and oversight and all those things. So I think this idea that we’re somehow going to forget everything we’ve learned for hundreds of years and just rely on an AI is not the right move.   Robert Bryant: Yeah.   KP Reddy: So I think what I would say is it’s all iterative. It’s not an all-in approach, but I think not doing anything is fatal. I think you have to-   Robert Bryant: Okay.   KP Reddy: … do something.   Robert Bryant: Thank you. Salla?   Salla Eckhardt: Yeah. My advice is that people need to tone down their ambition to be perfect. A lot of times when I talk about emerging technology to people, their first response is that, “I don’t know how to use the tools. I don’t know anything about this.” They forget that everything that they’ve learned thus far, they’ve done by learning. So doing their best and using the tools to their best capabilities and continuing to learn and having that open mindset rather than being fixated into, “This is how I’ve always done it, this is how it was taught to me way back when.”   Robert Bryant: Right.   Salla Eckhardt: That will open up their mindset a little bit. Even that much to be acceptable of AI as the supportive tool, and it’s not perfect. It’s never going to be perfect, nor are we as human beings going to be perfect, but allowing AI to be part of our everyday operations and allowing it to learn from us, but not thinking that we are the best at everything that we do. Being open into checking that how might AI actually get us to rethink about the processes and the ways of working that we are using that will tone down a little bit of the anxiety about new technology.   Robert Bryant: Okay, thank you. Sam?   Sam Zolfagharian: I just second that for the interest of time, I second that what KP and Salla mentioned so far. I believe educate your team and also experiment with AI solutions. Don’t set the expectation really high because it takes time. It’s like tuning your guitar, so it takes time to find the right tune for your guitar. So try to practice and experiment with AI solutions.   Robert Bryant: Fantastic. Just upon the screen here, Augment It. Sam’s book is available here. If you clicked through on here-   Sam Zolfagharian: Can I interrupt? Mehdi’s book. Sorry. Sorry to interrupt-   Robert Bryant: Sorry?   Sam Zolfagharian: Mehdi Nourbakhsh’s book.   Robert Bryant: Okay.   Sam Zolfagharian: Yeah. Sorry.   Robert Bryant: I’m sorry. I’m sorry. I know you facilitated this, so yes, if you mention InEight as you click through on the link there. Now we’re coming to an end, so I do need to bring it to a close, but I want to say a couple of things. First of all, a big thank you to all of you on our panels. Salla, Sam, KP, thank you so much. We could keep talking for another couple of hours I’m sure and keep exploring the theme. A reminder to our audience, please do take the survey. You will find that down at the bottom of the console. So please do complete the survey and also register for part two of this series as we explore AI in construction. So thank you very much to all three of you again. I wish we could keep talking. We will talk again, but for today, we’re going to have to wrap it up here. Thank you very much. Thank you all for joining. Thank you for your input. Hopefully, we’ve answered your questions, and please join us for part two when that comes up on November 8th. So for now, thank you and goodbye.   Salla Eckhardt: Thank you.   Sam Zolfagharian: Thank you so much, Rob.