Water Data Forum - AI and the Water Workforce

Water Data For - AI and the Water Workforce

[00:00:14] Max Herzog: Welcome, everyone. Thanks so much for taking the time  to participate in the last water data for session for 2024. My name is Max Herzog. I'm Deputy Director of Programs and Partnerships with Cleveland Water Alliance. And it's my pleasure to have you all participating in this series, which is put on by my organization as well as the Water Environment Federation and the Midwest Big Data Innovation Hub.

[00:00:40] Max Herzog: This is an ongoing series. We haven't yet scheduled our sessions for next year, but hopefully you've been tuned in with us for past sessions and we'll come back for future additions as well. , as I mentioned, I'm Max, and I'm with Clifton Water Alliance. We are a nonprofit. Economic development agency based in the Great Lakes region, and our focus is on accelerating water technology innovation.

[00:01:05] Max Herzog: For economic development and addressing water resource challenges. , and it's my great pleasure to be facilitating today's session. On AI and the water workforce with a really fantastic group of panelists. Experts in these areas for a variety of perspectives. , today we have with us Robert Bornhofer, who's director of innovation for DC water.

[00:01:30] Max Herzog: Miriam Hacker, who's the research program manager. At the Water Research Foundation. And Joe shuttle worth whose digital lead within water at a route. ,

[00:01:44] Max Herzog: I do see that we have some of their hands raised. I'm going to ask if you could perhaps lower your hand. I'm not sure that should be disabled. And so my apologies for the confusion there. We're going to ask that throughout the session and towards the end. that folks submit their questions , via text in the Q& A function.

[00:02:10] Max Herzog: You can find the Q& A in this sort of toolbar at the bottom of your screen. It is distinct from the chat, and so we ask that folks do submit their questions to the Q& A. That is where we will go first to, to look for questions at the end of the session, and we should have time to answer a few questions.

[00:02:33] Max Herzog: Panelists, you also have the option if you want and have the bandwidth as we're having the conversation to take a look at the Q&A and you can respond with text. But that's certainly not required. But yes, prefer participants encourage you to drop questions in the Q&A throughout or as we reach the end, you know, really up to you.

[00:02:52] Max Herzog: So with that, I'd like to just dive into today's discussion. , you know, we're tackling today. This topic of the profile of technologies around artificial intelligence and particularly their impact on work, the workforce in the water industry, water and wastewater. And so I'd like to just take some time to have our panelists sort of introduce themselves a little bit and particularly their perspectives on AI's use in an impact on the water workforce.

[00:03:23] Max Herzog: , and so perhaps Joe, we could start with you. I'd love to hear just a little bit about. you know, how you're seeing, how you've seen AI and are seeing AI used in impacting the water workforce.

[00:03:35] Joe Shuttleworth: Yeah, of course. Thanks, Max. , nice, to see everyone. Thanks for joining , today. So, yeah, Joe Shutterworth from Arup.

[00:03:43] Joe Shuttleworth: And my perspective is from, really the consultant lens. So Arup is a sort of, you know, 18, 20, 000 person engineering design consultancy. We work with lots of water utilities and really thinking about how we can solve interesting , I would say if you look back five or four years when we were first thinking about things like AI machine learning, you know, it was all really focused around, you know, data analytics.

[00:04:12] Joe Shuttleworth: So looking at time series data forecasting, you know, predictive maintenance, you know, flood forecasting, all those types of area. I think what's happened probably recently is outside trends hitting the water sector and changing how that paradigm is viewed with some of that being the emergence primarily of And a lot of the large language models.

[00:04:37] Joe Shuttleworth: So chat, GPT, and you know, all the work that's done there that really has taken away what people thought AI would be solving mathematical problems. It's now solving what are more creative problems and can help you with, yeah, accessing information, you know, the more mundane tasks. , and so probably how that's.

[00:04:57] Joe Shuttleworth: most relevant or that I would view to the workforce is probably in three areas. So probably thinking about, you know, AI as a coworker. So we've all heard of co pilot, but thinking about how we can use tools, as a partner in doing work. And if we look at outside sectors and how that's being done, you look at, you know, finance, you know, places like JP Morgan, looking at how to use co pilot to support them in making, you know, complex decision making.

[00:05:27] Joe Shuttleworth: So how could that impact people in the workforce when thinking about predictive maintenance, you know, water conservation strategies, things like that. Probably another area is, you know, skill building, and kind of, you know, not necessarily mentorship in a word, but really that sort of on the job training.

[00:05:43] Joe Shuttleworth: , so how can, those types of tools be used to provide access to information around. maintenance regimes around compliance, safety around operational protocols, you know, those types of things. And then probably finally automation. I think the next wave really of like transformation when it comes to digital data technology is around automation.

[00:06:05] Joe Shuttleworth: So how can we use you know, these AI based tools and approaches to automate processes that are done really back of house. So things like finance, procurement, reporting, things like that to unlock for the workforce. What are those more like creative higher level tasks? , that's just a couple of perspectives, you know, really interested to hear, hear what others think.

[00:06:31] Max Herzog: Thanks, Joe. Really appreciate that perspective. I think it's great to have, you know, sort of as an engineering firm, this sort of higher level overview of The landscape of technology adoption and implementation. I'm wondering, Robert, if we can turn to you, you know, as a utility that's very focused on, piloting and adopting innovation.

[00:06:51] Max Herzog: I'm wondering if you could share a little bit about your perspective on what a eyes use in an impact on the water workforces has looked like and looks like.

[00:07:01] Robert Bornhofen: Sure, and it's a delight to be here, Max, and to build on what Jill just mentioned. , so my perspective is one of the innovation of a, as an innovator, but from a WADI utility sector, we're already seeing, all of us are seeing how AI has already begun to demonstrate its enormous power and how work activities are being 

[00:07:27] Robert Bornhofen: Performed today, I think of a couple of vendor products. I'm not endorsing anyone, but we're seeing how vote AI, for example, is being able to predict likelihood of pipe failure using AI today. Okay. Redstone is doing the same thing with pipe. So we're seeing and I can go on, but we're seeing products already having impact.

[00:07:49] Robert Bornhofen: That are enabled by artificial intelligence and large language models. So it's, it's here right now. It's far more than just a technology upgrade. It's a fundamental shift and how utilities are and will be operating than in the past. It's, it's a bigger impact, I think, than the Internet. Now we have information.

[00:08:10] Robert Bornhofen: Now we can make decisions much better than before. So, and. DC Water, we're in a, we've been in pursuit of this, this opportunity with, starting with GPT 4 0, a year and a half ago, almost two years ago, and we're in the experimentation phase. , we're actually, Experimenting by building chatbots and we're also so we've got 2 or 3 applications of chatbots.

[00:08:40] Robert Bornhofen: , and they'll be used within the organization that help us save time. We're, we've adopted, we're a Microsoft shop. So we've adopted copilot. Joe mentioned copilot. We've got about 60 people who are subscribed to it. And, we're, we're learning, but we've got training starting off, three months worth of training through our Microsoft, trainer, soon.

[00:09:07] Robert Bornhofen: , we've been able to survey the preliminary results so far and using Copilot and on average we're saving about 10 minutes per application of Copilot like Copilot for Excel or Copilot for Word. It's actually helping us become more efficient. It's measurable. , and I think long term we're looking at, at our customer or our customer service area is a prime example, a prime opportunity where we can get some real savings and provide better, more, a different kind of customer service.

[00:09:39] Robert Bornhofen: 24 by seven by 365. We're looking at starting off with just doing a, a chat bot as we, as we, crawl and walk and start running towards a generative AI solution maybe a couple years out. We want to be able to provide a fully automated, customer care, customer experience with, with generative ai, multiple languages, natural language processing.

[00:10:06] Robert Bornhofen: We, realize that a vast majority of the questions asked can be. Answered. By a gen AI type solution. , and for customers who don't like that, they can opt out. And for a very complicated issues, they can opt out with a live agent. So we're, we've got kind of a tactical and a strategic view that is, it's, it's, it's evolving, but as we learn, we're adapting and we're looking at what else we can do.

[00:10:34] Robert Bornhofen: But from a, from our perspective at DC water and other utilities, this is a huge opportunity and it's one that we really need to take advantage of.

[00:10:45] Max Herzog: Thanks, Robert. It's really helpful to kind of dive deep on what some of these different applications are looking like from a utility perspective, both currently and forward looking.

[00:10:55] Max Herzog: , you know, Miriam, turning over to your perspective, I know you're coming at this kind of from the workforce perspective first and then thinking about the technology side of things. I'm wondering if you could speak to, you know, what you're seeing in the programs you work at at WEF and then just that general perspective of kind of looking at at the industry, from a consortion.

[00:11:16] Max Herzog: What your perspective is on A. I. S. Use in an impact on the water workforce.

[00:11:22] Miriam Hacker: Yeah, no, thanks, Max. And yeah, I will say we love collaborating with WEF. , I think that was one of the first starting points. My joined the Water Research Foundation was that there was such a passion for workforce. How do you put our heads together and figure out what we can do together?

[00:11:36] Miriam Hacker: , and so I really appreciate that relationship. One thing that I'll caveat on is the perspective I'm bringing into this is someone who kind of keeps an arsenal of research projects. So how do you understand the trends that are going on? How do you understand the questions that are being asked, in the research space?

[00:11:54] Miriam Hacker: And what is the research or the people that do the research that can answer those questions? And so the first thing that I always tell people is that what I speak on is not necessarily the research that I've done. We only do what we do at the foundation through our researchers that go forth with that expertise.

[00:12:10] Miriam Hacker: And so that's just a first big, thank you. , last night I was reading a report that was done just a few years ago in 2019 from Stantec, through the foundation that was talking about exactly, Joe, what you're talking about, that it's having a new coworker. It's simulating the training. It's. Optimizing your processes or automating them.

[00:12:31] Miriam Hacker: And one of the big findings was that, you know, it's not necessarily replacing people. It's shifting people. And I think what that highlights for us in the AI space is for all the technology that's coming into place. What is human wealth or the human  understanding? That either informs those decisions or the data that's being collected or maintains those systems.

[00:12:56] Miriam Hacker: And so what we had found in previous research projects was that people who, maybe are connected to AI tools or a digital transformation, you have to first understand what they know. And maybe it's upskilling a little bit to understand how to interpret the data that's coming out of these tools. Or maybe it's figuring out how to utilize these tools to help retain knowledge as people are moving through their careers.

[00:13:19] Miriam Hacker: And so I think for me, what this highlights is how much we need to reconnect and re engage with our, our workforce itself before we can implement these tools. Because the reality is we're collecting a lot of data. And to Robert's point, it's also thinking about customer service. As AI is more, familiar or becoming more popular in the water space, it also is in general society.

[00:13:42] Miriam Hacker: And how is that changing people's expectations for how they engage with the water sector? And so understanding how to bridge those gaps or how to connect with it in a meaningful way. I also throw out the caveat that People have challenges with resources. There was a question in the chat about what are the suggestions for small utilities?

[00:14:00] Miriam Hacker: That's a real concern. So how do we do more with less? How we do it more efficiently and how do we open up opportunities with what we have to do things better? And maybe a little bit more external facing. So I think that's kind of what I'm coming in thinking about is the people behind the tools. And it's a really exciting space.

[00:14:17] Miriam Hacker: I think there's a lot of opportunity there.

[00:14:22] Max Herzog: Thanks Miriam. And thanks, thanks all for this kind of first round. I think this does a great job of sort of painting a baseline picture of what this, what this landscape is looking like right now. You know, I think In general, when we hear about AI, there's, there's sort of this big polarization on the one hand, you know, AI is going to transform and change everything.

[00:14:42] Max Herzog: And on the other hand, like AI is, is overhyped. , or even on the third pole, perhaps that AI is dangerous and it's going to like damage everything somehow. So, You know, I'm wondering if we can just spend some time sort of, right sizing our expectations and kind of narrowing in on, you know, what, what is really the scope of what can happen right now and the risks associated with it.

[00:15:04] Max Herzog: And, so I'd like to dive into sort of y'all's perspectives on limitations, risks and barriers, that current limit. You know, the productive application of AI to water workforce workforce issues. Excuse me. , and I think Joe, I'd like to turn to you first with that question.

[00:15:23] Joe Shuttleworth: Yeah, of course. , I mean, I would think about it in probably three categories, in terms of limitations, risk and barriers, but really, Data,over promise.

[00:15:36] Joe Shuttleworth: , and then how workforce transition is managed. I think in the first bit thinking about data. , when you think about, you know, AI and machine learning, you know, it's Really, a lot of it's just, you know, it's very advanced statistics and the quality of what you get out of that is based directly on the quality of what goes into it.

[00:15:59] Joe Shuttleworth: I think people say garbage in garbage out. , and that is very true. I think at the same time, a lot of the success stories we hear are from very data rich. like consumer finance, consumer shopping and things like that. So you see applications that Amazon or Google or the big banks are using. , in the water sector, we don't have as much data as we think we have and where we do have it, it's either not as good as we think it is, or not in a format that can be easily understood by other people.

[00:16:33] Joe Shuttleworth: , we work across like a very wide range of sectors. So we do work in transport. We do work in buildings. We do work in water. We do work in energy. And when we're looking at applications of AI machine learning, they do vary across those different sectors in terms of both the like applicability and usefulness.

[00:16:52] Joe Shuttleworth: , and I think it is useful to look at different, you know, different sectors and what they're doing. You could take health as an example. So health. In on one sense, if you're looking at health, so using things like, you know, image recognition to detect, you know, medical diagnosis on scans and things like that.

[00:17:11] Joe Shuttleworth: Very promising application. But if you look at Data systems, often health has very fragmented data systems like water does, you know, even within a utility, you might have the customer function, you'll have the water supply function, you'll have, you know, storm water, wastewater, each maybe have different systems.

[00:17:32] Joe Shuttleworth: What the healthcare system did is develop, really like data interoperability frameworks. So how you, you know, Collecting and storing data in a way that is easy to access and transferable across different use cases. And so like a really nice thing to imagine is a, you know, a industry wide, standard that is adopted by the whole water sector.

[00:17:54] Joe Shuttleworth: to allow kind of integration across different asset management systems, billing, customer service platforms, things like that. , you know, the second thing is over promise, I would say. So the Gartner hype cycle is a really good image. People might've seen it. It's like the curve that goes up and then down, and there's the sort of.

[00:18:14] Joe Shuttleworth: I can't remember what it's all called, but there's the trough of disillusionment. Then there's the peak of expected inflated expectations at the beginning. And you place different types of technologies on that, on that cycle. And so you'd have things like, you know, hybrid analytics, like, you know, cloud, computing at one end or the, you know, realization and all these types of things there.

[00:18:38] Joe Shuttleworth: And I think that people are sometimes talking about AI machine learning as if it is a golden. a silver bullet, not a golden, not a golden bullet, a silver bullet. How many times have we seen presentations on digital twins at conferences that look like, you know, it's this sort of autonomous treatment plant or autonomous wastewater, you know, collection system, you know, that are we there yet, I think we have to be kind of honest with ourselves and reflect that there is work to do.

[00:19:05] Joe Shuttleworth: And I think that managing that is very important and thinking about where, you know, you know, AI is being used and what function it's replacing to focus on specific value to utilities or, you know, communities or, you know, whoever it is, rather than trying to, you know, claim to be a very universal solution.

[00:19:26] Joe Shuttleworth: I would say the third thing is around sort of workforce transition and I think that DC Water are a very good example of having a central innovation function. So where we're looking at these very like cross functional innovations like AI, like, you know, like, you know, big data, things like that.

[00:19:46] Joe Shuttleworth: Lots of Other companies like the big technology firms, like lots of energy groups have the central innovation teams that bring together people from I. T. From engineering from, back of house functions to kind of break down silos and make sure that that is brought forward. And, and the third, the sort of secondary bit around workforce transition is often the gatekeepers to technology adoption are people within the workforce supply chain.

[00:20:15] Joe Shuttleworth: So people that worry that their job is going to be replaced, but then also consultants, and I say that as a consultant, that are worried that the business model is going to be disrupted. By adoption of new technologies, so I'm not endorsing any products, but there are products out there that could do things like generative design.

[00:20:32] Joe Shuttleworth: Nowhere. You're automating the design process. You massively reduce the amount of hours required to do it. You know, consultant  isn't going to sort of bring that forward necessarily because they don't they don't have to. , and so thinking about how we kind of better manage that transition in a way, you know, where the benefits are shared across, you know, all aspects of our supply chain, I think is very, very critical in terms of, you know, addressing those limitations, risks and barriers.

[00:21:03] Max Herzog: Thanks, Joe. That's really helpful. I think you've given us. Yeah, as you said, like a couple of different really big themes to be exploring, Robert, I'm kind of wondering from your perspective, as Joe said, DC water has been ahead of the curve in general, in terms of, exploring and adopting innovation in the water sector.

[00:21:23] Max Herzog: , I'm curious. What limitations, risks, and barriers you feel like you've encountered in thinking about applying AI to the water workforce, you know, to what degree does it map onto these bigger trends Joe's speaking to, and, you know, more, more tactically, what's that look like for y'all?

[00:21:39] Robert Bornhofen: We could spend a long time on that topic, Max.

[00:21:41] Robert Bornhofen: , and I'll, Mariam, I'll save some for you as well. I'll cover three, I'll cover three areas. We're trying not to duplicate each other, but I, again, Joe was spot on with the data completeness, data quality, And the hype is a lot of hype right now. Okay, so I'm going to rule of three. So I think the first area, there was a study last, last spring by the Pew Research, P U G H, on, it surveyed, workforces, and it found out that, that people, employees, between the age of 18 and 30, about 32 percent are already using, GPT or a tool like that on the job.

[00:22:32] Robert Bornhofen: Okay. Many organizations, many organizations do not have, are not in front of this. It's being used and there's a lot of risk. There's a risk, there's risk of privacy. , there's risk of making wrong decisions, bad decisions. These, these so called, hallucinations. , but organizations need to get in front of this.

[00:22:55] Robert Bornhofen: They need to have an acceptable use policy to make sure that the employees are using it correctly. They need to be trained. There needs to be an audit plan to make sure that the risks are being mitigated. So that's one realization that utilities need to realize is that their employees are already using these tools.

[00:23:16] Robert Bornhofen: So you better get in front of that and put some guardrails in place to make sure they don't expose your organization to unnecessary risk. That's the first area. The second area of risk is, it is such a fast moving target right now. What, what I thought. What I knew a year ago has been replaced today. I mean, the rate of technology advancement is amazing.

[00:23:41] Robert Bornhofen: , we expect to see GPT 5 coming out soon. We expect to see better ability to be able to analyze an image, an image of a pump, for example, to be able to kind of understand what, what's going on. looking at the wear and tear. , we don't have that today, but we're getting closer and closer to having some even more amazing capabilities.

[00:24:05] Robert Bornhofen: So the fast moving is do you put a lot of money into it now, or are you going to replace your tech? For my example, with, customer care, customer service, We got to make sure that we're taking this in the right sequence, the right technology and the right migration plan. So fast rate of change. And, I think Jill touched on this, but it's the fear of someone losing their job.

[00:24:33] Robert Bornhofen: The fear of someone's sabotaging efforts to be able to advance this type of technology. And I think a lot of that is just, it's just fear mongering. They hear things on social media, and, and make no mistake, there will be activities. The mundane, simple activities that will be replaced. Okay. , this is nothing more than a tool.

[00:24:57] Robert Bornhofen: This AI, it's a tool to free you from those activities and be able to do other activities where you're using your mind more and more judgment. I've got this photo I love to show at presentations. It's like circa 1930. It's AT&T, and it shows At the time, at the time 19 shows 3 women in front of a switchboard and just kind of plugging the cord of the different phone numbers, the different sockets.

[00:25:24] Robert Bornhofen: And that's all they did all the time until we invented a automatic system for switch switching system. And that. Is the analogy of some of the routine things that we do over and over again, the customer care example, a customer service example of the mundane question as to how to open up a new account.

[00:25:44] Robert Bornhofen: I mean that's how many times you have to go through that spill. We can automate that we can listen, we can respond to questions. So I think the fear, the fear, the thought of the fear of losing one's job. I think we need to be able to have greater awareness, greater communication. In the realization that some activities will be replaced and that we do have career planning, we have career growth, we have places for those people to go to, but we, we don't, we don't have enough people in the water utility industry and we are all working hard, working long,  we could use this added productivity to make our jobs easier, more efficient.

[00:26:25] Robert Bornhofen: So it's welcomed. , and so I think that can be managed. And the third area, let's see, fast moving, fear, and then the, the acceptable and then having the fact that people are already using it, getting in front. So those are 3 areas I think are really key. , I think utilities need to take a look at that and manage that.

[00:26:46] Robert Bornhofen: So, Miriam, I left you a lot of room there for expansion on this.

[00:26:51] Miriam Hacker: Robert, you, you're bringing back memories. I'm pretty sure I've got a photo of my grandmother in front of a switchboard for, I think, a police department or something. So you're making me want to go back in the archives. Yeah, Max, just to kind of build on that.

[00:27:06] Miriam Hacker: There's three things I think touch on a little bit of what both Joe and Robert have talked about. And the first one is the data itself. And so I was working on a project this summer with a group of really brilliant graduate fellows looking at water data in the U. S. And the reality is it's really fragmented.

[00:27:25] Miriam Hacker: It's different qualities. The data management is really distinct. And I think that reflects the administration of data and just generally how we oversee our infrastructure in the country. It's pretty fragmented. It's pretty decentralized. The challenge is that everyone's affected. The challenge even more so is that it disproportionately affects smaller utilities.

[00:27:46] Miriam Hacker: So it then presents some sort of an environmental justice consideration of folks that don't have the resources to go out and find workforce folks that are able to do the data analysis, the data gathering and cleaning for utility might be less able to access, the benefits of AI than others. And so I think that's one thing that at a meta level, we need to think about as a country.

[00:28:09] Miriam Hacker: So, for example, this last spring, EPIC and the EPA just released the first publicly available map. Of water systems in the U. S. Just water utilities. I think they're currently working on the sewer systems next, which I'm super excited about. , but we don't necessarily even know holistically where things are situated.

[00:28:32] Miriam Hacker: And so I think if we're thinking about workforce, if we're thinking about utility progress and utility ability to do what it needs to do. Having access to that data or having access to it from people that maybe aren't able to go out and get that data themselves. I also think about the incremental steps that you can take to start moving towards.

[00:28:52] Miriam Hacker: AI. So there might be some leapfrogging that happens. The reality though is, are your records digitized to begin with? When I started my work in the water sector, I was working in a county in Washington state, and we were the digitization plan for our stormwater management systems. And so they had me reviewing And circling manually for all of our inspectors, all the aspects and components of our stormwater drainage systems, and then they would go out, confirm that was the case from the site plans, and then bring it back and so put it into GIS.

[00:29:25] Miriam Hacker: And so I got to work through that personally. So I think I realized, okay, there is a need before we can go to AI. How are we currently managing the data? And what is the data? And can it be improved? Can we improve the quality of the data before we start thinking about the different tools? Because like Joe said, you could have a really cool visualization dashboard, but the data going into it isn't very high quality.

[00:29:47] Miriam Hacker: It still isn't going to achieve the intention you had for it. , the second thing is the access to workforce has been really constrained to folks that are engineers. I have a background in civil engineering. I tell folks I tripped into social science and social things, so I'll get into that more. But , when we talk about who are we hiring, I'm starting to see different types of roles emerge, especially in communications, especially in public information officers.

[00:30:18] Miriam Hacker: We're looking for people from different backgrounds. There's starting to be more opportunities for people from public health backgrounds, and even thinking about AI. Where do people come in from different backgrounds? Like Robert said, we already have a deficit in the water sector for who's already here.

[00:30:34] Miriam Hacker: So are there ways that we can think differently? About this approach. I think it introduces this limitation. Like you had asked originally, there's a limitation on the paper ceiling who we are hiring for in the job descriptions were putting out there don't necessarily meet who could potentially be doing the job.

[00:30:51] Miriam Hacker: And so thinking creatively about how do we train folks that are non water into the water sector, I think, create some opportunities. I know, I know, people are thinking this. We know that the salaries don't match tech. We know that. But what I think we're seeing with generational differences, which is something that has consistently come up in our research since the 90s, there are generational differences.

[00:31:13] Miriam Hacker: I think we can leverage that when we're starting to think about these descriptions. Some of the younger generations really care about social good. So in some cases, they're not just going to go out for the money and the stability. They might be looking for the social good that's happening. And how do we tell that story for people to be excited and engaged in the water sector, specifically utility work?

[00:31:35] Miriam Hacker: The last thing that I would say is, you know, we think about AI, we think about getting a data management plan. We think about bringing in data analysts and the people who understand what to do with these tools. The reality is that the decisions that are being made about the tools, aren't necessarily going to be the people that use them.

[00:31:51] Miriam Hacker: And so that brings another conversation about working with leadership. This becomes a change management or an organizational culture issue of what does your strategic plan say? Is it aligned with that? Does your company feel like it's moving in that direction? Otherwise you might encounter issues like Robert talked about of the fear mongering or the resistance to change or worrying about your jobs.

[00:32:12] Miriam Hacker: And so again, when you're thinking about incorporating this, it has to be a larger conversation than that specific application of the tool and thinking about how do we move together. And there's a lot of work that's been done on organizational change and dynamics and transitions, and I think there's a lot to learn from it.

[00:32:28] Miriam Hacker: But I think sometimes. And like Joe had mentioned, there's a lot of hype around some of these tools, and you see this tool and you want to do it, and you're like, why can't we do that? And it might get shut down really quickly because there's some steps before that, filling out the waters, getting leadership on board, and making sure it makes sense for the entirety of the organization, not just for the specific application you're seeing it in.

[00:32:48] Miriam Hacker: And with some of these tools that are coming out, the integration across them can also be a challenge. Especially with maxed out  budgets, you can't afford to buy a bunch of different tools, or maybe you can, but if you can't, Are there ways that some of this AI can work better together? And so I think that also affects how we approach it.

[00:33:06] Miriam Hacker: So again, the degrees, how do we expand our thinking and change what's needed for the role? What are the absolute essentials and what can we train people for on the job if we know that we can utilize other strengths that they're bringing to the conversation?

[00:33:23] Max Herzog: Thanks so much, Miriam, and thanks everyone. This is really , helpful to get perspectives on sort of really the complexities of this space right now. I mean, we heard what I hear from everything you're saying is just incredibly sort of dynamic and evolving space with, you know, challenges with data quality and structure and gaps, you know, these really powerful social challenges of fear on the one hand and hype on the other hand.

[00:33:52] Max Herzog: The rapid rate of technology evolution challenges with the scaling down and thinking what it looks like for smaller utilities, as well as,  you know, the changing roles and employment needs for the workforce. So I think it really paints a picture of just. A lot of change right now, right? And, I think y'all spoke a little bit to some of how we move forward in a productive way in the face of that change lay in the groundwork in terms of, you know, tailoring expectations around the hype and the fear, getting structure in place for good data, quality interoperability, as well as, you know, kind of the ways that leadership and, you know, employees and the everyday workforce need to be You know, engaged and and, really engaged in conversation around this.

[00:34:39] Max Herzog: I'm wondering if we could look now sort of to wrap up the facilitated part of this, this panel. to the future and thinking about, you know, if we can, as, as this, this field matures, you know, to Joe's point, it gets more to that like stable position, in the hype curve. As we start to sort through some of these challenges, I'd love to get your perspective on what you see successful scaled implementation of, of the use of AI and the water workforce could look like.

[00:35:10] Max Herzog: What, what could that really enable for us? And, you know, also, if you want to speak to it, what do you see as long term risks? You know, if things really don't shake out the way that we want them to, we don't address some of these, these big challenges. , just would love to dive into that, that kind of future looking perspective as we wrap up here.

[00:35:28] Max Herzog: And I think, Robert, we can turn to you first to get your thoughts.

[00:35:33] Robert Bornhofen: Thank you. Yes, that's a great question. So, we can accept the fact that technology and advances as they get more and more advanced are here to stay and, and they're embraced. That's how we become more effective, more efficient, how we can do more to be able to provide a clean, sustainable product for our customers and also minimize the impact on the environment and make it more sustainable for our current future generations.

[00:36:05] Robert Bornhofen: Okay, big words, but, I see building on what's been shared so far. We'll get, again, data, data, data. We'll get, got that, we'll get that structured. We'll get that labeled in terms of what's secure versus what's general. But I see more and more specific, specific bots, specific, you know, very singular functional type commands.

[00:36:30] Robert Bornhofen: Or capabilities that we can start to build a catalog of these bots so we can begin to automate and be able to leverage this technology in a very reliable way. Okay, so more bots, better data. Okay. , for the workforce. , you know, I think the risk here for the workforce are among those who do not see.

[00:36:58] Robert Bornhofen: What this provides is a tool and resisting the resisting learning, resisting to take advantage of this tool. They may become obsolete, like, like Miriam's grandmother was in that photo with the AT&T operators. Okay, they have to, they have to learn, they have to adjust. It's no different than when the internet came on board.

[00:37:20] Robert Bornhofen: We learned to use the internet as a tool. So I think we're going to have to change the mindset, big picture macro, the impact,the risk, we're seeing more and more data centers needing more and more water to cool their, their systems, cooling towers. , I think about 70 percent of all internet traffic passes through northern Virginia.

[00:37:47] Robert Bornhofen: So you got a concentration of data centers, and it's exploding. It's exploding in growth right now. , you're seeing energy consumption. It's just exponential right now. The growth rate, Microsoft just bought, the, one of the functioning nuclear reactor at Three Mile Island that, that, that was never, never damaged.

[00:38:09] Robert Bornhofen: They're going to be using that. That's not, that was, that was, Was that? I think that was Microsoft. , no, it was Google. Google bought that. Microsoft now is building little micro nuclear power plants. To be able to have renewable energy to run their data center. So we're starting to see how industry is looking at, the scarcity of resources needed to continue that growth by looking at how water can be reused non potable water and also energy sources.

[00:38:37] Robert Bornhofen: So we can be able to grow with that. So we're saying, once again, we're saying how industry and how the workforce is adapting to these constraints. And being able to kind of figure their way out and be able to continue to provide a competitive edge, or in our case in the water sector, a product that our customers will appreciate and use for generations to come.

[00:39:04] Max Herzog: Thanks, Robert. Really appreciate you sharing that. , Miriam, turning to you, I wonder if you could speak a little bit to, yeah, just sort of what you see the future, the future of AI and its use and impact on the water workforce.

[00:39:19] Miriam Hacker: Yeah, I, you know, when I'm not doing workforce, I'm doing water reuse research.

[00:39:23] Miriam Hacker: So I get very excited, Robert, when you mentioned that, because it's, it's a hot topic. And when we're thinking about tools to incorporate, you're also thinking about what is your contribution to sustainable outcomes for the environment as well, because of the computational power that's needed through those data centers.

[00:39:39] Miriam Hacker: And so the water usage, the energy usage, I also think that points more broadly to just unintended consequences. And this is something we see. Beyond AI, beyond the digital transformation, whenever you have innovation, same thing with regulation, there's going to be a lot of opportunities, but there's also going to inevitably be some initial failure.

[00:40:00] Miriam Hacker: And so not getting hung up on that failure, but also some unintended consequences. So as much as you can try and anticipate what could go wrong with this, or maybe what could be created as a result, or just being nimble enough to respond, I do think in terms of opportunities, though, something that I found pretty interesting from working with, the University of Washington's eScience Institute is just the power of open source tools and figuring out how do we do more with less, how do we do things that are open access, or how do we find ways to help support, folks that may not be able to have their own in house, innovation team.

[00:40:41] Miriam Hacker: But still being able to access some of these benefits that we are seeing across the water sector. I also think that really opens up the door for partnerships. , I've seen it as a great recruiting tool. So research, but also AI and also some of these, Some of these digital transformation projects create a huge opportunity to bring in folks who may not have otherwise known that water was a potential career path and bring them in to experiment, test out their skill sets, and then bring them in for a long term career.

[00:41:12] Miriam Hacker: And so I think that's a huge asset. I think that's a huge opportunity that we have. And I just, there's a way for us to get creative. I think the emergence of AI speaks to that. We're getting very creative with what we do and using tools for benefits. But now how do we continue that creativity to start bringing in, opening up access to non engineers and making them feel like they're at home in the water space.

[00:41:37] Miriam Hacker: The reality is that we're all connected through water, which means that we all need folks. in the water solutions. It can't just be the engineers. And so how do we be intentional about that as we go forward? , and how do we create organizations that are able to do that? And I think with the utilities becoming more external facing, lots of opportunities and lots to think about.

[00:41:58] Miriam Hacker: So glad we've got, Robert on the call and DC Water represented, but yeah, a big opportunity space.

[00:42:07] Max Herzog: Thanks, Miriam. , yeah. And Joe, just to wrap up this portion, you know, let's speak a little bit to  your, your perspective on the future, I guess.

[00:42:18] Joe Shuttleworth: Yeah, sure. I mean, it's been a great conversation so far. And I'm going to be very honest, like, if we if we're going to look 20 years ahead, None of us have a clue.

[00:42:31] Joe Shuttleworth: You know, if you think, if you look back and think, 20 years ago, you know, did we have, you know, the, did we have our iPhones? Did we have, you know, video conferencing in the same way we do now? Were we using chat GPT? And could you have predicted that all of that was going to emerge? None of us probably would have done.

[00:42:49] Joe Shuttleworth: We make some predictions about the future, but they're going to be kind of, you know, off, you know, to some suitable extent. , I would say in in general, though, like I am optimistic about applications of technology within, you know, within our sector. And where could we go, you look at some of the kind of emerging applications around maintenance around like removing a lot of the more sort of bureaucratic tasks, you know, do you end up in a world where we have these more sustainable, you know, predictive self healing water systems, you know, I'm using, you know, like imaginative words kind of on on on purpose.

[00:43:27] Joe Shuttleworth: I would say that like in the short term, I think probably having some humility as a sector and realizing that we are not inventing these technologies, but we are adopting these technologies. You know, the big trends within, you know, any form of technology, you know, when it comes to things like AI and machine learning and those types of things are being adopted by, you know, from what the technology companies are doing or from what, and, you know, and then an energy companies, how they're applying it, or from what, you know, consumer, you know, consumer businesses are doing and having some kind of outward look on where.

[00:44:06] Joe Shuttleworth: they are and whether some of those sectors are ahead of us and then seeing what of those applications could be adopted into the sector might give us some kind of indications. And so I'd say from like a workforce perspective, you know, really thinking about what are the tools that are being developed by their it houses that, you know, all of the lots of utilities, you know, work with, say Microsoft, Google, you know, Amazon, you know, what are the tools they're and how can they really be applied, I think is How to look and kind of pick out what those sort of short term trends are.

[00:44:39] Joe Shuttleworth: I think the energy sector provides some, like some, some, some, some perspective or sort of near term perspective, you know, as a operator of kind of like linear utility, you know, as a utility operator of what some of the technology applications are there. I think in terms of risks, I would say that the downside risk in terms of our ability as a sector to.

[00:45:04] Joe Shuttleworth: provide to continue to provide the service that we do is fairly limited, because, you know, water is a, you know, water is a sort of base requirement of any functioning society or economy. And it's regulated to some extent, or was regulated to varying degrees of extent. And so there will always be a requirement to sort of continue to provide that.

[00:45:26] Joe Shuttleworth: However, how it impacts the workforce, I think is something that needs to be thought of carefully and how we manage that transition needs to be thought of carefully. If you look at agriculture and I'm using lots of examples from other sectors, kind of on purpose, but if you look at automation in agriculture, you know, it's massively increased production, but also displaced, you know, a large number of jobs.

[00:45:47] Joe Shuttleworth: If the water sector relies heavily on. You know, AI and, you know, machine learning and all these automation processes at the same time, we need to put in place safeguards, you know, around reskilling around balancing implementation to try and avoid some of those negative outcomes. , I think a key tool to doing this to managing the uncertainty is to think, and Miriam, you touched on it to think about.

[00:46:13] Joe Shuttleworth: Some of those, you know, more flexible approaches to decision making under uncertainty. So things like adaptive planning, you know, so how can we be setting out on thinking about identifying and adopting technologies without ending up with a load of technical debt with the flexibility to change our course as new technology emerges?

[00:46:33] Joe Shuttleworth: And so I think I'm I'm very optimistic about the potential of technology. But I think that the only thing we can really be certain about is uncertainty and how that's going to be developed over the medi to long term.

[00:46:48] Max Herzog: Thanks, Joe. I think that's a great way to wrap up sort of the facilitated portion of this, big perspective on, you know, uncertainty, but hope for the future and thinking about looking to other sectors for both innovation.

[00:47:02] Max Herzog: And then I think to Miriam's point. , you know, the source of talent that's required to really enable that adoption and scaling. , I do want to turn now for the last 10 minutes to some questions from the chat. , and I'll let you know anyone from the panel who feels moved to to chime in. For these, the 1st, when we've spoken to a little bit, but I'm curious if anyone would like to speak a little further about, suggestions regarding, enabling small utilities to engage with these tools, you know, folks that don't have, the resources or, you know, establish innovation practice like a DC water.

[00:47:44] Max Herzog: Do folks have any suggestions for how. How small utilities can really engage in this work.

[00:47:52] Robert Bornhofen: I have a couple of thoughts. , start small , train big, but, the price of a subscription to Microsoft copilot is 30 a month. Okay, per person. The cost of GPT 4 is, I think I'm paying a little bit less than that.

[00:48:11] Robert Bornhofen: It's in the 20s somewhere. It's not, you don't need to have a bazillion subscriptions, but start with a core group of people, your change agents, the people within your organization that are respected by others. And who have an appetite for learning and discovering and let them, let them provide some feedback, but start small.

[00:48:32] Robert Bornhofen: And the other point, and I'm looking at Miriam and the Water Research Foundation, they're funding a study that DC Water is part of to be able for the next 12 months to be able to do that for the industry. And that is, is to come up with an approach. How do you go How do you adopt, artificial intelligence, advanced AI?

[00:48:56] Robert Bornhofen: , what are the methods, best practices, maybe some sample GPTs? , but this is something that we're working on and we'll have a report published in October of next year. So I don't know if that's quite public yet, but it will soon be public, Mariam, but it's, and we can talk offline, but that's coming out soon.

[00:49:18] Robert Bornhofen: So the Water Research Foundation, the Water Environment Federation, and AWWA are all teamed up on that same need to be able to provide both small and large water utilities. With a roadmap, if you will, to start small and how do you how do you grow and how do you scale it,

[00:49:38] Miriam Hacker: Robert I think I saw Lisa McFadden on here.

[00:49:40] Miriam Hacker: Produced left had produced a cyber security. White paper, I believe. And so that's that seems like a really good resource. I saw a question there about cyber security and what to look at. What I would say to those points is that there's a number of resources that have been developed. And to Robert's point, there's ongoing research right now.

[00:49:59] Miriam Hacker: And so, Max, I'll send you an email with just a breakdown of some of those lists because we've done some Former projects. I think in 2019, we have two ongoing projects that are looking one at digital transformation in general, and two looking at implications for workforce and upscaling and retraining. , and so I think there's a lot of opportunities to begin with, of looking at what we've learned from others and to stay engaged with updates.

[00:50:24] Miriam Hacker: I know on our website, you're able to create a public plus account and then you can follow projects. So if that's of interest, I saw someone in the Q&A  asking about staying up Connected. , that's one way to do it.

[00:50:37] Max Herzog: Yeah, and I'm glad that you did mention cyber security, Miriam. I feel like my impression at least is that's another area where water can really learn from other sectors and that a lot of the solutions we really need to look at are the solutions that have already been adopted.

[00:50:49] Max Herzog: It's not that we need water cyber security solutions. It's that we need cyber security solutions. , to go down the list here of or sorry, Joe, did you want to chime in on this one?

[00:51:01] Joe Shuttleworth: , I mean, there are groups like there are a In California, it's the California Data Collaborative, I think it's, I think is the last say, and who are a really good group at sharing resources, like pulling resources about data and things like that, like data, data analytics, stuff like that.

[00:51:18] Joe Shuttleworth: And there are other groups like that, that have, you know, resources, I think Roberts, you know, identified some of the Microsoft tooling, you know, and the fact that it is very available and there is very, large amounts of support for it, I think really should be no barrier to anyone getting involved. All you need is a computer and the ability to, you know, search for the right things and look, look up information.

[00:51:39] Joe Shuttleworth: But a lot of the information about those types of tools is, you know, things that are trusted and built, you know, is, out there.

[00:51:48] Miriam Hacker: And Swan For. They're a great collaborative of being involved with these and a lot of really great resources coming out from there as well. , like, like the bird, swan.

[00:52:01] Joe Shuttleworth: Yeah, massively, sports one as well. Great. A really great group.

[00:52:08] Max Herzog: , to move forward to the next questions, like I think maybe the next two could be packaged together a little bit and we've spoken to it a bit already. , or actually a decent amount already, but thinking about. What it looks like to embed some of these AI tools just into everyday workflows, thinking about smart assistance, and things like chat GPT and, and what also comes along with those in terms of sort of reasonable guardrails around their use.

[00:52:36] Max Herzog: Any thoughts on those? On the panel,

[00:52:42] Joe Shuttleworth: I could give a couple initially, and then I'm sure Robert, you've got lots of examples. I think, you know, 1 of in the question, it talks about using things like city work in the field and how you might insert a smart assistant. And then there's. a little bit about, you know, ChatGPT and things like that.

[00:52:58] Joe Shuttleworth: I guess in some of the, some of where you're using some of the, where some of those tools are being used. So they may be there from, you know, CityWorks, or it's like from the Autodesk suite or something like that. They will have, or they will be developing integrations with some of these technologies.

[00:53:12] Joe Shuttleworth: So it's understanding how you can, you know, build on the tools that you're currently using based on things that the providers of those tools, you know, are adding on to it based on these technologies. Thanks. I think the second question around, you know, chat GPT and Copilot and things like that, you know, it's, very easy to integrate things like Copilot into Microsoft Word or Microsoft Excel or Microsoft PowerPoint to massively streamline your own personal workflows.

[00:53:39] Joe Shuttleworth: And I think that. in order to do it successfully. So we, do it internally for some of our reporting and things like that. You need to know where there are sort of like commonly repeatable tasks or structures and things like that, where you will find the kind of efficiencies. I think in terms of guardrails, there it's really about sort of data security.

[00:54:01] Joe Shuttleworth: So sometimes you're putting information into these things and it's using that information to then train the future models and whatever's going on there. Yeah. There are versions of all of these products that are, you know, bounded by your organization. So if you pay for an organizational subscription, your information isn't being used to then train the models and things like that.

[00:54:19] Joe Shuttleworth: , and so that's from a sort of like data security, sort of cyber security perspective from a guardrails in terms of like what people are doing. I think providing training from the organizational level around how these technologies should be applied and having clear policies.  and like stat, like operating processes, procedures and things like that is very important.

[00:54:41] Joe Shuttleworth: Otherwise, you'll end up with people using, you know, chatGPT and things to do all their work, which is obviously not a good thing because of hallucination and things like that.

[00:54:53] Max Herzog: Any other perspectives on this? I'm not sure if we'll have time to dive into another question, so it's okay if we Or we can try one more. Okay,

[00:55:02] Robert Bornhofen: if you have something.

[00:55:03] Max Herzog: Yeah, there's plenty of questions here. We definitely will get to all of them here. I think maybe a good last question here to tackle is just how do you get buy in on the information of these tools, particularly from the perspective of management?

[00:55:17] Max Herzog: Like, how do you build kind of that? Interest and moment towards innovation. You know, Robert, you talk about starting small. How do you get people toe to start there?

[00:55:27] Robert Bornhofen: No, that's a great question. And that's not easy. , it's , you'll quickly find you've got allies and you got people, you got leaders who are kind of skeptical of this.

[00:55:37] Robert Bornhofen: And I think it's just you can't underestimate the effort it takes to sit down, understand their concerns and be able to address those concerns. , whether it's research, whether it's let's do a pilot, let's test this, but it's you need to reign in your, your chief operating officer, your chief financial officer, your, your legal group, among others.

[00:56:01] Robert Bornhofen: And so it takes a lot of effort to get them engaged. And so, and I would say my biggest. Challenge, which we were able to come together on was with legal. They were very, very concerned about about privacy about leaking information out into the you know, the older version of GPT was using our queries are prompts to train its large language models.

[00:56:26] Robert Bornhofen: So there'd be data in there about DC water asking questions about something. And so we were very, very careful about that. And that's why we adopted an AI policy. That helped ease their concerns and it was reasonable. So again, it's a lot of effort, but it's necessary if you're serious about going ahead with this type of technology.

[00:56:48] Max Herzog: Thanks, Robert. , Miriam, I, we could have you speak to this, but we are at a time here.

[00:56:54] Miriam Hacker: No, no, no. Respect time. We're good.

[00:56:56] Max Herzog: Okay, well, I just want to take the time to thank you again panelists for You know, spending, spending, spending the time with us today, sharing your insights and thanks to everyone who attended, really appreciate this conversation and, the recording will be available for folks who want to, to check it out or dive deeper again.

[00:57:15] Max Herzog: , thanks so much. Y'all hope you have a good rest of your day.

‍