[00:00:00] Speaker A: AI can give you an answer, but it can't necessarily explain why things are the way that they are or understand the surrounding context. And so that's really where, again, when we think about the opportunities for AI and human intelligence to kind of be combined and work together, that's really where the optimism comes, as opposed to, hey, this AI is basically just a machine version of a human that's going to come in and do everything that we do and take all of our jobs away.
[00:00:28] Speaker B: I am Reese Tisdo. This is the future of water, which we talk about all the ways which companies, utilities and people are addressing the challenges and opportunities in water. This is episode 92, and I know it's going to be a good one. That's because I'm joined today by Bluefield senior research director Eric Bentler. I want to talk to Eric because he's been on the road talking about artificial intelligence in the construction sector.
This includes water and wastewater utilities. The public launch of OpenAI's chat GPT in November 2022 really set off a wave or tsunami of conversations about the impacts and applications of AI and related technologies across a range of industries and economic sectors. Sure, AI in many cases, it's a buzzword in many respects, but at the same time, and it's while grabbing the headlines over the last twelve to 18 months, there are real applications and impacts that are unfolding before us. This includes water and wastewater within. That includes jobs and includes processes. It includes efficiencies and just sometimes just operations of physical plant. So in the US utility construction sector, it's already facing chronic labor shortages amid historic levels of federal infrastructure investment. So AI can serve as a powerful tool for productivity, efficiency and safety. Like I said, Eric is going to talk a fair amount about that in the discussion I hope that I'm about to have with him. But in the wake of his presenting, you know, this all comes in the wake of his presenting to the national utility contractors association, Nuka, believe it was in Palm Springs about a week ago. So I wanted to tap his brain, see if he'd share some of those insights with you. So, and I can assure you there's lots of good stuff that will be discussed with, with Eric. So before we do that, I thought I'd share some news. Caught my attention this past week. The end of March, that is March 2024. Georgia lawmakers finally gave final passage to a change in state law that would allow a private utility to provide water services for new homes near Hyundai's upcoming electric vehicle and battery plant just west of Savannah. The measure now goes on to Governor Ryan Kemp to be signed into law. I think the expectation is that it will be signed into law. Hyundai had broken ground in October 2022 22 on its first us factory producing EV's and the batteries that power them. It's the largest economic development project in Georgia's history. Constructions move pretty quickly for the south korean automaker. So given the significance of it to the local economy, to the state as a whole. Supporters of the bill said legislation is needed to accelerate some of the home constructions that are affiliated or needed to house some of the workers in Bryant county, where the Hyundai plant is intended to employ 8500 workers for the seven and a half billion dollar plant, like I said, west of Savannah.
So as a result, Savannah water utility Management, which is a private company that supplies drinking water to 32,000 homes in 17 Georgia counties, has pushed to pass the bill. They're the ones who are going to own and operate this water treatment facility. So why do we care? Well, the passage of Georgia House Bill 1146 highlights really the complex interplay between state regulations, local government authority, and private sector involvement in water service provisions. So there's a lot to be said there. So one at Bluefield, we're always interested in state policies that are driving or inhibiting private participation in water. Does this open the door to a state that has been relatively quiet when it comes to investor and utility participation? Or is this really about, you know, is this a one off to support the Hyundai facility? So I think that is interesting. The other part of it is local government authority. So the local government authority, in this case the county, has been superseded or its authority has been superseded almost by the state who the state is really interested in economic development that benefits everybody within the state because of tax bases and jobs, et cetera. And it looks really great economically. But for how long is that going to play out? I mean, usually we're talking about competition for water in places like California, whether it be Southern California or Arizona or Colorado, Colorado reverbation, you know, competition for water among agriculture and industry and domestic household residences. So, and that's in the west, typically, but in the east, there is a real water problem. And so the local communities are concerned about this. Agriculture is a big base, but they also have their own water needs. And, you know, I know, like I said, when we look at the eastern US or the eastern seaboard, we typically don't think about water risk. But there is a real problem. It makes me think about a couple of years ago when Bluefield research did some analysis for the water Reuse association, where we did some regional profiles looking at different regions of the US. And what are the drivers and inhibitors to alternative water supplies such as recycling, treating wastewater. So what we would call water reuse and one of the bigger issues in the eastern seaboard in a place like Savannah or western savannah is they're using aquifers for their drinking water and water usage. But if it's overdrawn, as we've seen in other parts of, then there can be saltwater intrusion, the, which can impact the quality of the water supplies and create bigger problems. We've already seen that happening in Florida. We're seeing it happening in, that would be eastern Virginia, and we've also seen in places like Long island. So in New York. So it's a significant problem that raises a lot of issues when it comes to water management. So, therefore, have not seen the designs of the Hyundai facility and what they're, you know, I know how much they're drawing down or expected to use. And they've got permits for four different wells in an adjacent county and they're working on that. But you know what? We don't want this to be as a tragedy of the commons where everybody gets access to it. The local authority is overlooked in its insights. I thought that was interesting news. So, like I said, role of private participation in Georgia, interesting competition for water among different stakeholders in Georgia or the rest of the US. Definitely very interesting applications for new technologies, alternative water supplies, also very interesting. All coming together just west of Savannah. So if you're interested in this, let us know, and one of our analysts can always jump on a call to talk to you about it. So with that being said, we'll get to Eric and talk a little bit about AI and what's happening in the construction industry.
All right, I'm joined here by Eric Benller. Eric, welcome back. How's it going?
[00:07:40] Speaker A: It's good. It's been a, been a busy week, but looking forward to the weekend. How about you?
[00:07:44] Speaker B: Pretty good. I'm not going to. I mean, we were just talking about some interesting conversations we've been having about sort of the state of the world elections, the water wastewater sector. So, yeah, I'm, for everybody to know, it's Friday, so this is it for us. But this is it. Exactly. So why don't we, why don't we get into it? So is in the intro, I let everybody know that obviously, I was talking to you, but also the fact that you have been looking at artificial intelligence, its role in the construction industry, but also water wastewater and what all of this noise, real or unreal, might mean. So what I was hoping to do is have you kind of share some thoughts from some presentations you've been making, but also some of your research as well. So I feel like, you know, like every day I open the newspaper, which I still do, I'm old school, I do open the paper, paper and I read about AI and all the disruption that's happening. And since the rollout of chat GPT water sectors also either jumping on the bandwagon or thinking about getting on the bandwagon. But its everything from angst to opportunism when we look at it. So can you maybe put some of this buzz in perspective why the water wastewater sector is talking about it?
[00:09:09] Speaker A: Yeah, its definitely been a topic of conversation in a lot of places. I mean, just a couple months ago, I was at the utility management conference in Portland, and id say at least every panel slot, if there were maybe five or six panels going on concurrently, at least one of them was about AI or digital data more broadly, but especially AI. It was a really hot topic of conversation. Companies are talking about it, utilities are talking about it, investors are talking about it.
I had the opportunity to talk about it myself at the NUCA conference in Palm Springs a couple of weeks ago. That's the National Utility Contractors association. More of a construction market crowd. A lot of this is actually more geared towards utility contractors and construction firms as opposed to utility operators themselves. And so it was a bit of a different crowd and a bit of a different kind of a topic for me to think about from kind of a digital water perspective, but really interesting nonetheless. And one of the first things I wanted to do for my slides, I was asked basically to give a presentation on the state of AI and kind of applications of AI in the construction sector. The first thing was kind of, how do we put some sort of numbers or framework around understanding how big this thing is, right? I mean, again, it is, there's a lot of buzz around it, especially since the emergence of chat GPT back in late 2022. And so it was kind of like you google how big is the AI market or how fast is it growing, and you're going to see all kinds of crazy projections and stats out there and estimates. I think there was some, I dont remember if it was Deloitte or somebody else that had put out estimates for AI is going to add 7% to us GDP. These really big numbers that are at this point, I think, very hard to quantify. But what I wanted to do was look at what weve seen in terms of growth so far, looking at actual reality and the past instead of projections.
Two interesting stats that I thought were powerful and helpful for this were basically, were this. So the first one was looking at the time it's taken different technologies and different platforms to reach their 1st 1 million users.
You can find a couple of different versions of this type of data out there. But Netflix, for example, when it launched the streaming platform in 1999, took three and a half years to reach a million users. Facebook, when it launched in 2004, took ten months to reach a million users. The iPhone in 2007 took about 74 days. And chat GPT, which is really the first really mainstream AI platform or application out there, five days to reach a million users. And so really, really quick growth in terms of just the spread and the adoption to date. Now, again, it's a free platform. It's not necessarily dollars spent, but it does give you some indication as to just how quickly this thing has taken off. Our colleague Amber did a blog, I think, last year on the impact of artificial intelligence on water and data center spending and things like that. And she put in there, she had found us that. So, yeah, ChatGpt launched November 2022, a million users within five days, by August of 2023. So less than a year later, ChatGpt had 1.4 billion site visits in the month of August 2023 alone. Right. So were talking like pretty rapid acceleration of kind of growth and of usage. The other interesting number that I found was in kind of the venture funding space, thinking about just the investment thats going into this market broadly, obviously not just in the water sector, but kind of across the economy. And so that was that in 2023, about a quarter, more than a quarter of all venture funding for us startups went to AI based startups. Right. And the percentage for the prior five years was something more like, you know, ten to twelve, maybe 15%. So again, really, really rapidly scaling investor interest in this technology. And then if you know, you kind of compare that to the public markets and kind of the rise of Nvidia over the past several months, right, this company that makes processors for AI has skyrocketed to become the third most valuable company in the world, behind Apple and Microsoft. So to give you a sense of the public investor side of it and how much money is going into this, how many dollars are being put behind this. So all of that is to say, yes, there is a lot of hype around this, but to some extent, theres some pretty significant numbers in growth and acceleration to justify that. Hype. The last piece ill throw out as much as I was just saying, you know, that we shouldn't think about the forecast and we should think about what's actually happened. We, of course, have our own view as far as AI applications within the water sector. And so this was another kind of visualization that I put together for the NUCA presentation. 2024 market size us water wastewater, capex and opex $177 billion growing at about a 2% compound annual growth rate throughout the rest of the decade. Right. So pretty slow. A large market, but not very fast moving. The digital water market within that, about $8 billion. So a pretty small sliver of the total pie, growing at about an 8% compound annual growth rate. And then weve also tried to just identify AI powered or AI enabled technology segments and products within the digital water market.
And thats something like $400 million. So pretty small sliver of even just the digital water market, let alone the entire water wastewater market, growing at about a 14% compound annual growth rate. Right. So that gives you a sense of the scale here. It is still a very small piece of overall activity in the water wastewater utility sector, but a 14% growth rate equates to something like a doubling every four to five years. Right. And so that level of change and growth is pretty uncommon for the water industry. And I think it makes a lot of people uncomfortable. Theres a lot of implications for jobs, which well get into. And again, so its, it's a real phenomenon. It's still small, but it is real. It is growing quickly. And it is important to talk about whether you're a utility or a contractor or a vendor that's serving the industry.
[00:15:30] Speaker B: Yeah. Like I said, from angst, optimism, or opportunism, there are opportunities because of this.
As you were mentioning, you were presenting at NUCA.
When most people think of construction, they're thinking about pick shovels, bulldozers, pushing dirt and laying water, wastewater pipe in the ground. But AI didn't immediately come to mind when they're thinking of that. In fact, up until now, they haven't thought about it probably at all. But to get everyone on the same page, can you define AI, maybe even like, what is it and maybe what is it not?
[00:16:09] Speaker A: Yeah, I mean, this was really how I kind of kicked things off at Nuka. This was really part of what I was asked to do is just, you know, this is a crowd that you're absolutely right, is, is usually thinking more about pictured shovels and bulldozers and not necessarily, uh, you know, artificial intelligence. The most kind of cutting edge technology. So it was just a level set. You know, can we put some basic kind of, you know, user friendly definitions out there? And so found a couple ways of thinking about this that I thought were helpful. I mean, the first point to mention is that, you know, AI itself, artificial intelligence is, it's kind of an umbrella term, right? It refers to a range of different techniques and technologies that have manifested themselves in different ways, in different products that people are using or in different solutions. So I'll just define, there's quite a few out there, but I'll define a couple that I think are the most maybe commonplace or the most relevant to this conversation. And the first would just be kind of AI itself. I mean, actually, this definition, kind of stealing it from the, we actually put out a white paper on AI in the water sector way back in 2019, I think, with Arcadis, so pretty early days in terms of this discussion. So shout out to Jim Cooper from Arcadis for this one. But his way of saying this is basically artificial intelligence is the ability of machines to sense reason, engage, and learn in a manner that seems intelligent, which I think is a really succinct and powerful definition there. Within that, you get to things like machine learning. Thats another really common subset of AI that basically involves enabling machines to learn and improve based on data or experience without being explicitly programmed in terms of what to do. And so that's kind of the classic example of essentially trying to teach a machine to identify a cat, a picture of a cat, not by telling it what a cat is, but by giving it a bunch of pictures of cats and letting it kind of work out what the common features are and how a cat is different from a dog and things like that. So just feeding the algorithm, feeding the machine with enough data that it's, it learns the patterns to identify one thing versus another thing.
Then generative AI is the third key term here that is really coming to the fore with the chat GPT and the OpenAI conversation. And generative AI basically is artificial intelligence that uses machine learning to create brand new content that can be text, it can be code, it can be images, it can be videos, it can even be music in response to a user generated prompt, something like a chat GPT. Basically, it's trained on just massive, massive, massive amounts of text based data from the open Internet to be able to produce human like text. You can ask it questions, but it's not just about giving you the answer, finding the answer, but creating text in the format that you ask it. So you can ask it to write you an essay or a term paper. You can ask it to write you some lines of code. You can ask it to write a poem. Actually, one of the guys in the NUCA audience was talking about how he had used it to write a love poem for his wife. Right.
You can have this. It understands not just content, but also kind of style and tone, which is really interesting. And so that's kind of where we're at with this. The Chechi Bt conversation is that piece of it, the generative AI, kind of creating new content. So that's kind of what AI is. I think the other piece of it that I was really interested in talking about was kind of what AI isn't, right. And really, a lot of that comes down to. And the place that I wanted to take that conversation is the difference between artificial intelligence and human intelligence. Right.
The understanding at this point is that we're kind of in a place with AI, with kind of the overall development of AI. It's called narrow artificial intelligence, right? So people have been successful in designing specific AI programs or applications or algorithms to do very specific things, right? Whether it be something like a chat GPT or, you know, a treatment plan optimization algorithm or what have you. We're not at a place where there's an AI program that. That has kind of reached something that would be, like, on. On par with human intelligence and especially the. The diverse set of skill sets and. And. And knowledge and kind of functions that the human mind or the human brain can perform. Right? That would be something called, you know, artificial general intelligence. And we're still, you know, years, if not decades, away from that. The next step, the thing that's really kind of the scary, you know, Sci-FI futuristic thing would be artificial superintelligence, where AI is actually kind of smarter than or more powerful than the human brain. But there are still these really important differences between what AI is good at and what humans are good at. And that really gets to the conversation that we'll get into in a minute about the workforce implications. But AI can process really large datasets really quickly. It's really good at identifying patterns and trends, making predictions, spotting anomalies or discrepancies in these massive amounts of data that individual human would just not be able to process or make sense of in any meaningful way. And then there's the absence of human error or fatigue that you get with people, right? You can ask an AI, you can ask an algorithm to do these really complex calculations, to do these repetitive tasks, and it's not going to need a break. It's not going to need to go to sleep. It's not going to make the same kind of human errors that, that a human operator might be able to make. There's also the opportunity for really, because AI can work at such a great scale, the opportunity for really personalized output. If you think about the classic Spotify or YouTube, where essentially by processing all of the data on the decisions that you've made within that platform, the songs you've listened to or the songs you've skipped, for example, and comparing your behavior with millions of other users that are doing the same thing, it can make recommendations about, oh, if you like this artist, you might like this artist and get down to that really personalized level in a way that would be really tough for an individual person to do. But there's a big difference between that and what the human mind is maybe best suited for and the strengths that humans still have far and above what AI is even close to being able to do. And that would be in areas like empathy and emotional intelligence, creativity, being able to make decisions based on things like ethics and norms and morals, and even just understanding common sense. AI can give you an answer, but it can't necessarily explain why things are the way that they are or understand the surrounding context. And so that's really where, again, when we think about the opportunities for AI and human intelligence to kind of be combined and work together, that's really where the optimism comes, as opposed to, hey, this AI is basically just a machine version of a human that's going to come in and do everything that we do and take all of our jobs away.
[00:23:04] Speaker B: Trey, I mean, I can tell you what it shouldn't be doing is writing love poems to your wife.
[00:23:09] Speaker A: Yeah, sounds like you got in trouble for that one, but it was a pretty funny story for them.
[00:23:14] Speaker B: I don't know if I'd announce that to a crowd, let alone my family or my wife, but good grief. Well, with that being said, and super helpful, because there is a lot to unpack. I mean, one of the big controversies, or at least that everybody talks about when it comes to AI, is its potential impact on jobs. Right? So you've come across some really interesting data on this and applied it to construction and, or we've started talking about water, wastewater. So can you provide some color on this and what it might mean for types of jobs impacted and so on?
[00:23:55] Speaker A: Yeah, I found this really interesting analysis, actually from Goldman Sachs or some economists at Goldman Sachs on. Yeah, basically just trying to understand what the potential impact of AI could be on different types of occupations. And so I, you know, unfortunately since it's a podcast, I can't kind of show you the slide, but if you're a client, definitely come check it out. We've got the Nuka slides on our website. I'm happy to kind of share this one around because it's super interesting. But basically the idea is that they looked at, I think, 900 different occupations, like not necessarily jobs, right? So it's not in terms of number of people, but it's 900 different occupations that people have in the US economy. And they tried to model out what percentage of that person or that occupations kind of daily workload, what their tasks associated with that occupation. They tried to figure out what percentage of that workload could be offset or exposed to AI in some way. Right? So if you're a market researcher and you spend part of your day answering client emails and part of your day doing research into new topics, which of those tasks could potentially be taken over by an AI? And how does that vary for different types of jobs? A market researcher or a construction worker or an investment banker? So what they found was that basically it's about the first 40 or so percent of occupations, the kind of least exposure occupations, basically have really no or very minimal exposure to AI. Then you get into this kind of middle range of about 57% of occupations where something like a quarter to half of the daily workload could be offset by AI. So a significant portion but not the majority. And that's where you get into these conversations about AI being able to free up the time that you're spending on some of these more mundane, more manual, more data related tasks and allowing for some of the stuff that's more human, right? The creativity, the critical thinking, the kind of emotional and social intelligence interacting with people, the stuff that maybe is higher value in a lot of contexts and a lot of businesses.
That's the interesting conversation there. And then really the top 5% is the piece that's maybe a little bit in trouble, right? That's where you're seeing about more than half of the workload potentially being offset or automated by AI. Those are the types of positions where, yeah, there may actually be some job losses or people needing to kind of be reskilled and retooled and move into new types of occupations. But I thought that was a really interesting way of thinking about it. That again, for the biggest chunk of occupations, again, 57% right in the middle, there will be some exposure, theres some opportunity for productivity gains and for freeing up some of that time and bandwidth to potentially do more higher value, to spend more time on the other 50% to 75%, thats maybe more valuable, thats more creative and thats more human and relies more on human intelligence. So that's kind of the top line. And then when you translate that to construction, it's really interesting, right. Because really, construction is very much on the lower end of the scale for the most part. And so here I basically looked at kind of, I think it's about twelve or so, twelve to 14 or 15 kind of high level, like Bureau of Labor statistics classifications for jobs within the construction industry. And actually this, this one was specific to utility construction as opposed to other types of construction. And so as you can imagine, the vast majority of people that work in the construction industry are construction workers, right? So people that are involved in construction or extraction activities, it's like about 50 something percent of the total workforce. That's a pretty low exposure to AI. There are some interesting applications that we'll talk about for improving worker safety and things like that, but we're not anywhere near a point where AI is going to build a building or build a utility pipe or build a water treatment plant, right? We're not there. The next big chunk is installation and maintenance and repair type jobs.
That is another probably 15% to 20% of the utility construction sector in terms of workers. And that's also pretty low exposure. Again, similarly, an AI is not going to be able to maintain or repair a pipe or a treatment plant, but then you start to get into the back office type functions, office and administrative support, management, business and financial operations, architecture and engineering sales. Those types of roles actually do have quite a bit of higher exposure to AI. And so that's where you're seeing a lot of the potential applications here. It's not necessarily using AI on job sites, although there is a bit of that, but it's much more using AI for expediting or facilitating or improving the scheduling process, the estimating process, the design process, the project management stuff. That's really where a lot of the opportunities are and I think speaks to, yeah, just the different types of skill sets that are really part of the utility construction sector.
[00:28:57] Speaker B: So, yeah, I mean, I think, I mean, as you talk through this, I mean, I'm sort of thinking about, you know, those can work from home and those cannot, you know, if you work from home, there's a high likelihood that you are exposed or we are exposed. Whereas if you're, you know, if you're digging ditches, someone's got to do it. There's no one.
There's no AI that can really go do that. And so skilled craftsmen and such, maybe in the design and engineering, that's where applications can obviously take place. But within the construction industry, I know you were meeting with some of these people. Are they using AI?
What's the uptake of artificial intelligence within the construction sector?
[00:29:45] Speaker A: Yeah, so this is a really interesting piece of it as well. It's digging around for just some state of the industry survey type data to get a feel for what are the big issues that the construction industry in the US is facing. Utility construction, but also just more broadly. And when you map it out, a lot of it is stuff that AI could potentially help offset. So I've got, again, another great chart here that if you're a client, definitely go on and check this one out. But the top three concerns for construction firms right now are all financial and economic, right? It's interest rates, it's trucking or insurance costs. It's the prospect of an economic slowdown. That's not really anything that AI could do anything about at this point. We're not at a place where AI can prevent a recession or drive down interest rates or inflation, but basically everything else is pretty interesting to think about in the context of technologies like AI. So the next three big concerns are labor costs, labor quality, labor supply. We already talked a lot about just now about the potential workforce impacts there, but also stuff like material costs and supply chain delays, permitting and inspection delays, education and training regulations, policy impacts. And so as I was digging around trying to understand where construction workers are using AI, it touches on a lot of that. And I think we'll spend some time kind of digging into some of those more specific applications. But AI to again help to offset some of those labor shortages by making workers safer, making workers more productive, especially at the back office level, expediting things like compliance reviews and things like that. Helping with optimizing supply chain and purchasing decisions in a really uncertain supply chain environment. Again, using AI to process massive amounts of data and help make those decisions. Or helping to create more realistic schedules and, and cost estimations based on past project results and things like that to reduce the friction and the delays and the unexpected costs that would come later on in the process. So a lot of ways that AI can help construction firms across the industry. But on the flip side, when you look at how much it's actually being used, it's still pretty small portion of the industry, of the construction sector that's actually using this stuff. So in the same kind of survey that I looked at, there was actually a question on how many construction firms are using AI, have made investments in AI and about 70% have made no investments per this survey data. Right. 70% havent done it at all. 19% have kind of made their first initial pilot scale investments and then another 11%. Or maybe at the next step theyre stepping up their investments. Theyve maybe made it beyond the first initial pilot stages. So the dichotomy here or the case here being that theres a lot of opportunities for construction firms and construction workers to benefit from this technology, but the vast majority of construction firms are not yet using it. Now, I will say one of the points that was observations that was made when I gave this at Nuka, actually by the same guy who had written the love poem to his wife.
He was a talkative guy, but he mentioned that part of that may just be the fact that of those 70% of firms that arent using AI, they may just not even realize it yet. They may be, they may be using technology and they may be using software that has some sort of a basis in AI and artificial intelligence and they don't even realize that. Or they're not aware or maybe somebody in their organization is kind of playing with it, but it's not like a company wide initiative. So I think that's a really interesting angle and I think would probably be pretty translatable to the utility sector as well, where if you go out and ask a bunch of utility managers or executives or CEO's, they may say, no, we're not doing that, but there may actually be somebody somewhere in the organization that is, or they've invested in some software platform that has some kind of underpinnings in AI that they just don't realize. Right. So that's an interesting point. But the last getting into the reasons for this, and I think this is really where a lot of what my talk was focused on was. So okay, we see that there's a lot of potential for AI within the construction industry, but most firms aren't using it, or at least not at any kind of meaningful scale. So the question is why? What are the barriers to adoption? And this is where I found actually some pretty interesting Deloitte research on not in construction specifically, but just kind of across the economy, just top barriers to AI adoption from global Global CEO's, a survey of Global CEO's and the big barriers are not really related to cost in any way. It's all kind of about it's all basically knowledge and information gaps. Right. The biggest barrier is just identifying use cases. It's business leaders not really maybe being aware of this technology, but not really understanding how they can use it or what's the appropriate place to use it or where to start trying it out. But it's also things like just overall digital maturity or capabilities or lack of data, just, again, not having the in house capabilities, the in house expertise, the in house kind of data to support any kind of scalable AI investments or strategy. And so that was really kind of an interesting observation that I think would also be pretty well translatable to water utilities more specifically, or to construction firms more specifically. Is that the big challenge here is that people know it's out there, they just don't really know what to do with it or how to make good use of it. And that's really where I was trying to focus the presentation. Clay?
[00:35:18] Speaker B: Yeah, I think it's really, I mean, it's simply like, what is the definition of it and what does it mean and what are the applications? Because artificial intelligence is a sort of umbrella term. I think the other thing that was interesting, and I agree with, I mean, it got to 1 million users in five days because chat GPT was super accessible. Right? It's free. You can go on, you choose it, you use it, you poke around. I mean, I did the same thing. I was probably one of those 1 million people. So with that being the case, it is now, I suspect, now with not just chat GPT, but other Claude is another one. Depending on your organization or just depending on your interest, there are a number of AI tools out there that are accessible to all of us. But you also see, like LinkedIn even says, hey, we have artificial intelligence. Can we help you write your posts? They do all these things. So I think it's really interesting. It is out there, and whether there's a true difference between whether it's been formally introduced and is being managed within a company or organization versus not, or is it more free form, more ad hoc among the employee base when you look, let's maybe quickly go through some of the applications where you see some companies using AI. And do you have any potential business outcomes for AI in the construction sector?
[00:36:42] Speaker A: Yeah, absolutely. And so this was kind of fun for me because, I mean, to your point, being one of the early users of, like, using GPT in the first couple of days, I actually haven't done much with it, but I figured this was a great chance to work with it.
Part of my day job at Bluefield is really understanding really well the water utility technology and software landscape. And so I could very easily list off a bunch of names of startups that are doing this stuff in the utility space. But I hadnt really looked at it from the perspective of the construction industry, software firms and startups and AI companies that are selling into general contractors and construction firms. So I thought, well, what's a better, you know, what better place to start than using chat GPT to come up with some lists of names of like some of the, some of the big possible use cases and then who's actually doing that or who's kind of developing solutions for that. And you know, I think it was a pretty telling experience. You know, we actually, one of our colleagues was just talking to a sewer based AI company a couple of weeks ago that does basically cctv inspection, you know, using kind of machine learning or machine vision on that footage. And the remarks from her, from the CEO of the company was basically like, this is kind of like an intern, right? It's something that can save you a ton of time, but it's not going to be 100% right all the time. You do still need some sort of checks and balances and kind of reviews and QAQC in place. And so that was really what I found here. I asked it to give me a bunch of company names, and maybe 50% of them were pretty spot on. And others either had gone out of business since, you know, basically Chet GPT, the free version of it at least the information is only good up until about 2022 because that's when the training was done. So anything that's happened after that, it's not going to know about. So companies went out of business, or companies got bought by somebody else, or they were rebranded, renamed, or they just did AI, but not at all in the construction industry. Like, nothing to do with construction. So kind of an interesting thought experiment for me in terms of how useful this is for market research at least. But I did come up with some pretty good names and some pretty good examples of applications. And like I said, a lot of it is in the kind of the pre construction or the back office space. So it's, you know, using AI to optimize the design process. There is actually a cool company that you can basically like, it's kind of like a chat GPT, but for designing buildings, you can kind of tell it what you're looking for. You can like type out, I want to, I'm going to build a building that has these characteristics and, and it'll like, make models for you. Right. It'll. It'll do kind of design preliminary designs of, of what that, that building could be and help kind of expedite that process. There's a lot of stuff, as I mentioned earlier, for reviewing past project data and current economic, environmental conditions to say, what is an optimized or realistic cost estimation for this project or scheduling for this project. A lot of it comes down to basically document review. So, feeding complex legal documents into an AI platform and then basically being able to ask questions of your PDF's. Right. If you've got a couple hundred page legal documents or contract documents and you want to say, what are the main provisions of this contract? What are my major risks here and my liabilities here? Does my design for this project meet local code and compliance standards and things like that? Being able to use technology to help with that really manual document review and document management, I thought was super interesting. Um, and then you get into the construction side of it, which there's a lot of really interesting stuff, too. I mean, I know I said earlier, right, that more of the potential exposure is on the back office side of things, but there are some pretty cool technologies that are, you know, maybe not necessarily offsetting construction workers, you know, core duties, but, but, but are kind of helping with things like efficiency and productivity and, and really, safety was the big one, right. So there's some companies out there that do things like basically processing, kind of monitoring video footage from a construction site and checking to see is everybody wearing a hard hat, and if not, send an alert to the foreman or to the supervisor so that they can come in and fix that. Or I've even heard about things like tracking people's eye activity, like people behind the wheel of a bulldozer, for example, or some sort of a vehicle, tracking their kind of eye activity to make sure that they're not falling asleep and stuff like that. So just really interesting applications that are, again, that's not going to take over for a construction worker, but it is going to potentially make them safer and prevent accidents and prevent injuries and liability and lawsuits and that kind of thing. There is still a little bit of activity out there that's actually kind of like robotics infused with AI, I should say. So probably a lot of people have seen that Boston dynamics, that company that makes pretty advanced robots, and they have this little robot dog that can run around a job site and do basically collect surveying data and that kind of thing. And there were some interest cases, interesting examples of autonomous bulldozers or autonomous pile drivers and things like that. So we are getting to that point as well. I think that's definitely far, far earlier stage than some of the stuff that we're seeing in terms of the back office applications and the other piece that I've talked about. But that is out there too. And it's really interesting to watch how that unfolds in terms of the outcomes. This was another interesting question. Okay, so theres all this cool technology. What can be done with it? And so again, looking through some work from companies like Deloitte, for example, some of the estimates that I saw range from in terms of kind of ten to 20% reduction in budget or timeline deviations from over the course of the project, savings on engineering time spent or engineering hours of like ten to 30%, and then total project cost savings of ten to 15%. Right. So pretty significant numbers, especially in a pretty tight margin occupation or tight margin industry, that I think would be really meaningful for a lot of companies. The other piece of it that we havent talked about in terms of outcomes and opportunities that I think it was just kind of one slide that I threw in randomly in the presentation because I thought it was important and this is something weve talked about quite a bit. The other side of it being for all of this AI to function, there needs to be a lot of data centers out there, a lot of data processing centers, and all of the construction that we're seeing going into that space. Just looking at the US facility count data, there's been about a 15% compound annual growth rate from 2015 through 2023. And just the number of data centers in the US, there's just a tremendous surge in building activity here. And again, if you're a utility contractor, utility construction worker, there's a ton of infrastructure that's going into building those facilities. Right. The water supply for cooling, all of the energy supply, the telecommunications connectivity infrastructure that needs to go into all of that. And not only are there more and more of these facilities, but especially because of the AI boom, they're getting bigger and bigger. I just saw a couple weeks ago an announcement from Oracle. They're building a new data center to house a lot of their new AI related workloads. And they said you could basically fit eight Boeing planes nose to tail in this facility. It's that big. Imagine again, all of the infrastructure, all of the utilities that are part of that.
So not only is there an opportunity specifically for this audience, for the utility construction audience, it's not just using this technology, but also being able to help build it and contribute to that infrastructure that I think is a really important and interesting opportunity.
[00:44:28] Speaker B: Well, and the water usage for those data centers, I think our colleague Amber has looked at this about. For every thousand chat, GPT searches, or 50,000, how much water does it use? I mean, it is. The water demand from these, for cooling for these data centers is immense and growing. And currently, you know, I know they're working on all kinds of solutions, everything from putting data centers, you know, submersing them in the ocean and, you know, cooler waters to a number of different things. So it's pretty incredible. So. Well, most of our research, I mean, a lot of time you spend actually on the digital space that focus on utilities rather than construction firms and general contractors. So having gone through this exercise, talked to a number of different companies at NUCA, UMC and elsewhere. Do you have any perspective on similarities or applications as far as digital maturity for AI and or water wastewater?
[00:45:34] Speaker A: Yeah, I mean, that was interesting for sure. This was really the first time that I've spent in kind of a dedicated event focused on the construction sector and on construction firms, not on utilities. Right. And so one of the things that struck me was just there are, I think, a lot of similarities in the sense that the kind of contractor space is also very fragmented, right. It's very regionalized and localized. You've got a lot of small firms that are operating in specific areas, just like you've got a lot of most utilities are kind of small municipal operating in a particular city or a particular community. Right. So I thought that kind of fragmentation was really interesting. And with that, you get a lot of the same concerns or challenges related to just broader technology adoption. Right? If you're talking about a small firm, maybe a generally older workforce, there's maybe not as much digital maturity, there's not as much time or interest or understanding or budget for data and digital technology and things like AI, right? So there are actually a number of kind of construction software firms. They are exhibiting at the, at the event, and they all kind of. A lot of what they were saying about the challenges that they face in terms of demand and adoption really resonated with what I hear from digital water companies that are selling to utilities, right. That there is a barrier there to overcome in terms of driving digital adoption, in terms of increasing digital maturity in this sector.
On the other hand, construction firms are private companies. Theyre for profit companies. And so there is that profit motive there. And a lot of the applications really were quite different in terms of how you're using these technologies. I mean, basically, a lot of the utility applications are for sustaining assets and operations as opposed to the construction phase. The job site applications, really more project specific. So using AI to reduce costs or improve compliance for a one off project basis versus something that's going to be used to maintain your operations over time, whether that be reducing your chemical usage or improving your asset management operations and investments and things like that.
Definitely some similarities in terms of the overall tech environment, but in terms of the specific types of companies and specific types of solutions, it was quite different. And I thought that was really interesting just to think about those two different phases of infrastructure assets, the firms that are actually designing and building them, versus the organizations that are then managing them and operating them over the long haul, and what that means in terms of where AI can be used and what types of benefits can be realized from using that technology. A lot of differences there that I hadn't really thought about.
[00:48:24] Speaker B: Yeah, I think it'll be interesting to see how it unfolds, whether it be water, wastewater, but also construction industry. I mean, pretty conservative. It's not to say that there's not innovation happening, but they need to get things done and do it well because there's potential fallout on the back end if it's not done well. That's everything from building a facility and making sure it's done well to things like water quality, et cetera.
So AI, we haven't even gotten into cybersecurity. We can talk about that another day. I mean, this potentially even opens the door, or opens the door to a bigger conversation about cybersecurity and what all of that means. So this is super interesting. I know to your point, the conference you attended was really interesting, love letters and all, which I can't get over, probably won't anytime soon. But with that being said, super helpful. Before I let you go, as I always ask, what else are you working on?
[00:49:24] Speaker A: So as you mentioned at the top, we've had a lot of great client conversations, client briefings lately. We just earlier today did a deep dive into potential implications of the US election for the water sector. And I think we're going to do something with those insights. We were just talking about that beforehand. So definitely a really interesting thought provoking conversation around what some different scenarios might look like. Weve also been doing some stuff pretty frequently doing different types of briefings and conversations with clients around digital water, whether that be global trends, us market trends, meter market trends, whatever the case may be. So a lot going on there as far as working with and supporting our clients. But then on the report side, working on stormwater, getting pretty close to wrapping up our first ever us stormwater Infrastructure market research report. If youre a client of our us muni service, thatll definitely be coming your way soon. Looking also at the industrial water market data centers, as well as food and BeV manufacturing, semiconductor manufacturing, mining, oil and gas. A lot of interesting activity happening on the industrial front. And so Amber Walsh, who weve mentioned a couple of times already, is working on a dedicated report, looking at that space. And then on the digital front, were going to be doing a market share exercise of the top 20 global digital water players and kind of what they're doing and how they're doing it, their strategies and their kind of overall position in the market. So a lot of super interesting, super relevant stuff that clients have been asking about that we've been kind of working on for some time. So looking forward to getting all that out the door.
[00:51:03] Speaker B: Yeah, it has been busy, there's no doubt about that. So appreciate you jumping on for this, and I appreciate the work and effort you put into not only presenting at the NUCA conference, but also sharing some of these insights. And I think to your point, when it comes to the discussion we had today, I think we've already posted or going to post your presentation on our website. So if you are a client, you should be getting access to that in due time, if you haven't already. So. Well, Eric, it's Friday. Like I said, we kicked it off. This is it. This is the end, almost. So I hope you have a great weekend and we will talk again soon.
[00:51:47] Speaker A: Sounds good, Reese. Thanks a lot. Have a good weekend.
[00:51:49] Speaker B: All right, man. Cheers. Take it easy.
All right. There you have it. So that was awesome to have Eric join on.
Not going to lie, we were making some jokes about sometimes AI isn't always correct and is one of his points made. I think it was. As he said, a scottish soccer team was trying to use artificial intelligence to track the soccer ball and where it was going on the field. And one of the linesmen actually walking up and down the line was bald. And so the AI, for some reason, focused in on the linesman with the bald head rather than the soccer ball. So I think, to Eric's point while laughing, was that AI is not always right. So keep that in mind when you write your love poem. So before we sign off, if you're in Boston, Barcelona, let us know and we'd enjoy the opportunity for a meeting. Please subscribe to Future Water podcast and give us a review. We're getting more reviews. It's super helpful. And if you're interested in anything we talked about today, from AI to election impacts on the water wastewater infrastructure sector, send us a note at water expertsluefieldresearch.com. And also throw out a topic idea if you're interested. And lastly, tell a friend about it.
This podcast and these water industry insights have been brought to you by the one and only Bluefield research. To learn more about us, visit
[email protected].
