The Public Works Nerds

GIS and GeoAI in the world of Public Works with Adam Carnow and John Shain

September 12, 2023 Marc Culver, PE Season 1 Episode 17
The Public Works Nerds
GIS and GeoAI in the world of Public Works with Adam Carnow and John Shain
Show Notes Transcript Chapter Markers

In our 17th episode we talk to two GIS Nerds, Adam Carnow from ESRI and John Shain from Bolton & Menk. In the spirit of the AI movement, which we talk about in this episode, specifically GeoAI, I'm going to let the AI generated episode description say the rest - I usually consider this too wordy to use for the actual episode descriptions and end up writing a more succinct version, but we're going to let you see what the AI machine generated for this episode. Enjoy!

From AI:
Prepare for a riveting journey through the world of GIS as we sit down with industry heavyweights, Adam Carnow from Esri and John Shain from Bolton & Menk. With over five decades of combined experience, these mavens of geospatial analysis provide an illuminating discussion on the disruptive influence of AI in their domain. Discover how AI and GIS are redefining public works, from managing sign installations to tracking homelessness, and explore the far-reaching implications of these advancements for the next 25 years. 

Unearth the advantages of cloud-based GIS solutions as Adam and John delve into Esri's innovative solutions developed in collaboration with agencies. Learn about the remarkable ease-of-use of these tools, their plug-and-play features, and the limitless possibilities customization offers. Understand the crucial role of user feedback in making these solutions more robust and efficient. Also, unveil the profound impact of AI applications in GIS, as we dig into machine learning and deep learning models and their impact in public works. 

The latter part of our conversation takes you through the fascinating world of real-time data and predictive modeling in disaster preparedness. Take a peek into successful AI implementations like Charlotte, North Carolina Water Department's innovative use of cameras for locating water meters. We also reflect on the evolution of Public Works over the past five decades, acknowledging Esri's instrumental role in propelling AI technology forward. This episode promises to be a captivating discussion on GIS, AI, and their transformative impact on public works. Don't miss out!

Show notes:
I wanted to highlight something that Adam said during the podcast because it highlights GIS's prevalence in Public Works, and government operations as a whole:

"GIS is a mission-critical enterprise business system that supports the full range of public works activities."

Six Focus Areas:
·        Operations & Maintenance
·        Streets, Roads, & Bridges
·        Capital Project Design & Engineering
·        Field & Fleet Management
·        Public Grounds & Facilities
·        Emergency Response

Additional links:
APWA Reporter – AI: Friend or Foe? https://apwa.partica.online/reporter/july-2023/marketplace/ai-friend-or-foe

APWA NC Newsletter – Why You Need to Know About GeoAI for Public Works
https://northcarolina.apwa.org/wp-content/uploads/sites/41/2023/06/APWA-Newsletter-2023-June_6-15-23.pdf

NACo Blog: Everything You Need to Know About Artificial Intelligence & GIS
https://www.naco.org/blog/everything-you-need-know-about-artificial-intelligence-and-gis

County Innovates Using GeoAI to Inventory ADA Curb Ramps and Saving Significant Time and Money 
https://www.esri.com/en-us/lg/industry/public-works/stories/county-innovates-using-geoai-to-inventory-ada-curb-ramps-saving-significant-time-money

Adam Carnow:

Welcome to the Public Works Nerds podcast.

Marc Culver:

Welcome to the Public Works Nerds podcast, a Public Works podcast of the nerds by the nerds and for the nerds. I'm your host, Marc Culver. Thank you for joining us today. Today, we're joined by two GIS nerds Adam Carnow from Esri and John Shane from Bolton & Menk. Let's start with Adam. Adam has been with Esri for over 15 years and has been working in the GIS world for over 30 years. Adam, why don't you just take a couple of minutes and talk about your background and your experience?

Adam Carnow:

Sure, and big, huge thanks for the opportunity to join and support your podcast and participate in it. I love the whole context and style of it, so happy to support it. Yeah, coming to you live from the Esri Regional Office in Charlotte, north Carolina, and, as you mentioned, I've been with Esri for 15 years. I'm in my third position at Esri. Started out as a local government account manager serving the southeast here in Charlotte and throughout the southeast region. I was a community evangelist for a few years and then a couple of years ago, switched over to our industry solutions team, which is industry focused marketing, and I'm focused on the public works industry. Prior to that, I worked 15 years in AEC firms three different ones in the Florida area, grew up in Florida, went to the University of Florida, have a degree in geography and another one.

Marc Culver:

I'm not a gator.

Adam Carnow:

I'm not a gator, I just said that for Adam's sake, and I have a master's in urban regional planning, so that's who I am and how I got to be where I am today.

Marc Culver:

Great, great Well. Thank you for joining us and looking forward to this conversation. Now let's meet John Shane. And John is a Ben with Bolton Mink. He's a superstar here at Bolton Mink for almost well. Okay, you just had your time, had your 24th anniversary with Bolton Mink.

John Shain:

Yep, I just flipped the calendar.

Marc Culver:

So next year I'll get to buy you a drink and celebrate your silver anniversary. I'll welcome that. All right, all right.

John Shain:

That'd be great. Put a great calendar.

Marc Culver:

Yeah, but you are the GIS work group leader at BMI, and so how did you get into GIS?

John Shain:

Luckily I was just lucky, I guess. Really, going to college I didn't really know what I wanted to do, like many, many prospective students. Yeah, I started out going down the computer science route. My parents gave me a good bit of advice in the late 90s that follow technology, follow computers, which I did. I got in going in computer science and it was okay. It was okay. Yeah, you know, the first couple of computer science classes can be a little bit dry, yeah, but at the same time of course I was, you know, looking in a bunch of different avenues and stumbled into geography in a 101 class. Of course they started to introduce some of the things you could do with geography. Course one is teaching. I knew I wasn't cut out to teach, although I'd done quite a bit of teaching as the years have went on. But they started talking about GIS and being able to use, essentially apply some of this computer science that I was interested in to something that was a little bit more visual, right, and that really sent me down that path and I took as many technical courses as I could at the time. So that's how I stumbled into GIS, yeah, and that was like the late 90s. That was late 90s yeah, late 90s. And actually when I first started looking in a GIS it was really kind of mid-90s right. So at that particular time when I first started, I was into command line, unix-based ARC info, yeah. And when I finally got into classes that you know talked about ARC view, it was the 3X series 3.1. It was just like this is it? This is what I was really interested in, so cool. That's where it started.

Adam Carnow:

Yeah, I'm in the same era, though I started working well a little bit before you. I started working with GIS in 1991 in grad school and it was all command line on UNIX workstations and then it went up from there, Right, See you.

Marc Culver:

And wow, look at where we are today. Could you guys even have imagined? I think both of us are shaking our head in a new way.

John Shain:

Yeah, no way.

Marc Culver:

Yeah, so let's jump into that then. So throughout the previous 15 episodes of this young podcast that I got going on here, I've heard so many of my guests talk about GIS and how it's transformed the public works industry. So today we're going to talk about that and we're going to talk about how we see GIS being used in the everyday life of public works, where we're helping our clients John, even Adam where you're helping your clients tap into that tool. And then we're going to move into the future and, spoiler alert, the future is here and I think I'm really excited to talk about how artificial intelligence AI is already a disruptive force in this industry, but how that's going to continue and really where we see this going. And like back in the mid, you know, the late 90s, we never could have imagined where we would be in 25 years. So who knows what this is going to be in the next 25 years? But let's speculate on that a little bit about where this is going. So let's dive in. We're going to assume that our nerd community if you're listening to this, you're obviously a nerd knows what GIS is, but you know geographic information systems. We're going to start with an assumed knowledge level that you know, you've seen it, you've used it and if you've used any, you know city or county mapping interface. Well, that's a version of GIS. It's using GIS information, but, but let's, john, how do you see GIS being used in public works realm right now? How are we helping our clients, you know, with GIS?

John Shain:

Well, I would say this that, first of all, gis you know, especially the Esri products in my in my opinion, has become foundational to the IT stack, to the technology stack for almost every community and, and that that being said, that it's much more accessible than it ever was. You know, I'm not going to keep talking about 25 years ago, but when we talked about that one big computer that had all the dust on the top that one or two people could maybe operate, those days are gone. Right Now, everybody has a device in their hands, they have multiple devices at home, working remote, in the office, and everything is delivered via the web and all of these devices that we carry around. So, you know, the types of applications and technologies that we're delivering people in a very targeted way has also changed, right. So the way that we're using, the way that we're analyzing data is is completely different, but it's so much simpler and so much easier. Now I would say that, you know, for a long time, there was just a lot of data collection that was going on. Right, communities and organizations were trying to collect all of the assets that they had, whether it was infrastructure or anything else. Right, those were being done with means of GPS and surveying methods, but now a lot of that data is collected for communities, and now it's we're in the mode that we're maintaining this information but also enriching the information, enriching the data with, with, with new capabilities, and maybe we're looking at the data in a different way. So what's been really interesting to see now is how all of these folks that maybe don't classify themselves as maybe technical, gis or even computer users are now walking around and they're performing analysis and data collection on the fly, every single day, you know, with these units that are all connected out to the same data Sources and it's. It's really amazing to see.

Marc Culver:

Yeah, yeah. So you know, just kind of getting back into the specific applications for for public works. You know, maybe talk about how and we're going to have multiple episodes If this, as this podcast grows, we'll have multiple episodes on GIS but we're also going to have multiple episodes on asset management. So talk about maybe a little bit and both you know Adam jump into but talk about how you know GIS has become foundational for for asset management as well.

John Shain:

Yeah Well, I mean, from what I've seen, gis now becomes essentially that hub where all asset management applications are reaching into, and GIS in its own can do a very nice job with asset management. However, what we've seen is, with maybe some of the other technologies that are leveraging that geospatial component, is there able to tie in things like work orders and get a really good idea of what the cost is and maintaining an asset through its life cycle, right? So we've seen web services being absolutely transformational there as far as being able to integrate GIS into their platforms and, like I said, there's a lot of good solutions that are just out of the box with being able to do, you know, relational databases and tables and things of that nature to be able to track activities. But some asset management applications are absolutely taking it to the next level as far as being able to do preventative maintenance and tracking those types of things throughout its process.

Marc Culver:

Yeah, and Adam, I know Ezri's got a lot of partners out there that are, that are in that asset management realm. What have you seen? That's really been impressive in that asset management world where your partners are really pushing the limits of, or the advantages of, gis.

Adam Carnow:

Well, before I get to that, I'd like to start first with talking about the kind of the scope of GIS within public works.

Marc Culver:

Yeah absolutely.

Adam Carnow:

I'd like to lay that kind of context down before we get into asset management. So the first thing to understand is that GIS is a mission critical enterprise business system that can support the full range of public works activities.

Marc Culver:

Yeah, I think that's a really great way to say it.

Adam Carnow:

Yeah, too many people assume that it's great for field work or it's great for asset management, or it's great for fleet or whatever, and they use it in one specific area or workflow, when really it is designed to support everything that public works does, and public works is a spatial business. I mean everything it does has to do with location. So if you're not utilizing that location, you're leaving a heck of a lot of ROI in that data that you're paying a lot of money and time and resources towards the manage. So and this is evident right from APWI I mean, every year since 2018, apwi surveys its members and they have like an NCAA basketball bracket to identify the top five tech trends for every year, and every year, gis has been one of those five top tech trends. So the industry is saying, yes, we get the value of GIS and it is foundational to our business. So that's great. But when you come at this from as a GIS professional, which I am, you can't. You must put the technology into the realm of the public works professional. And so at Esri, we try to echo what APWA does, and so we look at the application of GIS to public works into six focus areas. So those focus areas are operations and maintenance streets, roads and bridges, capital projects, design and engineering field and fleet management, public grounds and facilities and then emergency response. So those are the six kind of buckets where we kind of talk about the application of GIS. And so if you look at asset management, primarily asset management will fall under operations and maintenance, but it will trickle across a lot of the rest of those other five as well, and so the importance of location and asset management is just can't be understated, right? I mean, if you get a recall notice from a manufacturer for a certain piece of pipe and you've got to replace all those, how do you know where they are? Right? I mean, you don't want to start digging up and trying to find them and not know where they are. Or if a disaster comes through and wipes out a bunch of infrastructure in a specific area, how do you know what was there? Right? I mean, you've got to look. You've got to have locations as part of asset management. So we've got a lot of strong partners. You know Esri has over 3,000 partners across the globe in all industries. We've got a bunch that do asset management at different levels, whether it's enterprise asset management or work order management. All of that together We've got some top notch partners in that space and they're all taking advantage of location. But you're right, john, in that you don't have to go out and purchase a massive system if you don't want to. There are innate out of the box capabilities that RGIS has where you can start to build asset management capabilities on your own very inexpensively, very quickly, with little risk on having to code and throw out some apps to do inspection, to do data collection, to do some rudimentary analysis to get going. So really, you know we've had, you know, city, towns, you know 4,000 population that are utilizing our tool to do asset management in some shape or form.

Marc Culver:

Yeah, and you know, way back in the early stages of my career at the City of Maple Grove here in Minnesota, you know we were managing our first real asset management experience holistically was science and we were using our GIS you know, our map for managing our science, and there were new requirements coming in, you know, from the Feds on retro reflectivity and that, and so we were trying to get a handle on what we had out there and we had a really talented GIS analyst who was able to build us a little, you know system where our sign maintenance guy could see himself on the map and know what sign he was next to, you know, and just start using that like that. Well, we eventually Maple Grove, eventually, you know did buy one of the big asset management systems and they're using it for a lot more. But that was, that was the beginning, that was kind of the start, and they use they just were using Esri and I'm sure the tools have gotten even better since then and there's a lot more flexibility in that if you do, if you do just kind of want to build your own system there. So so that's. I think that's really impressive. Maybe talk about some of those other I don't know if they're templates or other things that you know users could use just kind of out of the box with with ARC map right now, you know.

Adam Carnow:

Well, not with ARC map. No, that's dating Sorry. Yeah we know we don't say ARC map around here much anymore. It's our GIS, pro Our GIS. I'm sorry. Right right right. So our ARC map is being sunset and I know we have a lot of happy users with that, but time moves on and it's time for everybody to move on the ARC GIS pro. But yeah, so very important thing that people should know if there are GIS customers and users is about what we call our ARC GIS solutions. So this is a collection of solutions that has redeveloped with agencies. So we don't develop them in some back room, you know, on a whiteboard going. I think we need to do this. We actually work with cities and counties and state agencies and really regional agencies to develop these hand in hand, using real data and real workflows every day, and we develop these applications and then we put them out there Think of these as an app store and they're all free and we have over 150 of them and there's about 50 of them. So a third of them are public works. But don't just look. If you go there, you can go to solutionsgiscom, you can look at the collection of them. Don't look at just the public works ones, because some of the ones outside of public works will be applicable. But, for instance, let's say, you need to do sign management. Right, we've got a sign management solution and you can literally check a box and say deploy this in my ARC GIS online cloud organization and it develops all of the products needed. You then start collecting data with the forms that are already set up. You can configure the form to change them if you need to. You go out, collect the data. It goes right into the system. There's dashboards ready to go. Everything's all set up. So literally, you know, in a day you could have a sign management solution up and running and testing it out in the field, and you know getting the ROI. So the risk is negligible. If you do want to download it, you can deploy it on your own. You know on-prem system if you want to, but there's really very little risk. It's supported by tech support. When we come out with new versions, we're going to upgrade those so that they will keep working. So it makes your GIS much more sustainable. You don't have to constantly be editing code and trying to keep up with all the patches and everything else. It's just plug-and-play and off you go. That's great.

Marc Culver:

John, what solutions have we used or implemented for clients I?

John Shain:

would say those solution templates have been wildly successful, especially in the ArcGIS online environment, and you know we've deployed several of them for clients everything from, you know, tracking homelessness and the unsheltered in place, all the way to doing citizen problem reporting, where we're engaging the public to be able to communicate where they're noticing where there might be some sort of maintenance issues. But the ones that we seem to be doing most often recently, besides citizen problem reporter, are, of course, right now the lead service line sort of solution, and that's right now a very you know, it's a very current topic where we're trying to identify where all these are, and the great thing about the solutions is not only does it give those folks that are managing the data tools at their disposal, but what I love about them is they also engage the public right. There's always a public-facing component, and one of the biggest challenges that we had through the years, when you're starting to put these applications and platforms together, is how do you also make this information available to the public and be able to scale it for many people to be able to use at the same time, right? So that's one of the things that we've been able to deploy a lot with the ArcGIS online environment. That has just been fantastic. So you know, and Adam has kind of said, you have flexibility there because you might deploy something and 20 minutes later you have an application, but you might have certain business needs and you're still able to readily change it. So it's not necessarily just a black box, right. You are able to customize it to your own needs, use what you need, discard maybe what you don't, and be able to deploy something that you know truly works for your organization.

Marc Culver:

Is there a back and forth on that then, like with these solutions if you know it's been out there for a little while and what agency like really starts with this solution but then evolves it into something else, is there then an opportunity to put that evolved advanced solution back into the mix for everybody else?

Adam Carnow:

Yeah, the team that runs that you know that program is very open with the community and they embrace the community and they want feedback. So, yes, if there's something that the solution doesn't do, that you want added to it, or you add it and you find value in it, give the feedback back to them and they review it and if they think it's right, then they'll put it into the next one. Or if you have an idea for a brand new solution, that doesn't even exist. So it is really a community driven program and the team goes out of their way to solicit feedback from the users of those solutions. So, yes, by all means, we want the user community to guide the development of those solutions, and so we constantly come out with improved versions all the time. I think they have quarterly releases and so recently, like last year with a big release, one was we had improved our winter weather solutions. There was a big upgrade there. That's important up in your Minneapolis region and we'd worked with Syracuse on that, who is, you know, they compete for the golden snowball right of the snowy city in America every year. So, yeah, so definitely, we want feedback on those solutions, make them better, make new ones, etc. And so that's great, john, you brought up the homeless ones. Not a lot of people know about that one. The lead service line is certainly a big one, but one that I'll also throw out there. That won't come up if you say you know on the gallery, only show me public works ones Is we have one for social equity analysis. So with this funding coming from the feds around a lot of infrastructure, some of the requirements for that money is that you make sure that it's equitably used across the community, and so we have a social equity analysis solution that will allow you to put a social equity lens on any discipline so you could apply the public works infrastructure investments and make sure it is equitable. So you can get this federal fund.

Marc Culver:

You know, I think, as we talk about some of these Different solutions in that you touched on the homeless Solution and the social equity Analysis and that it sparks the thought in my mind that you know we're talking about GIS for public works. But I know in both of the cities that I worked for we had really talented GIS staff and of course, at some point that GIS staff start supporting other departments within the city with their needs, because it is such a fantastic and Flexible tool that obviously I mean obviously land use planning and you know so there's so many other Applications and solutions for community development and economic development and things like that. But then even beyond that, for police, for fire, I mean it's, it's really it touches every aspect of city operations and helping, you know, produce some reports and analysis and and Assisting with with operations and planning. So that's just something to note. That a lot of public works people end up Becoming experts in some other realms within the city, just just helping them out and hopefully you know they've got the resources to get their own GIS staff as well.

John Shain:

But typically a snowball effect within cities. When it's when it, when it goes, it goes really well.

Adam Carnow:

Yeah, yeah. Well, I mean a lot of these problems that the cities and counties and agencies are dealing with Cut across departments anyway. I mean, homeless is not just a problem in public work, it's a problem with public health, it's problem and planning, it's a problem in Housing, I mean you name it. It cuts across everything law enforcement, I mean code enforcement, all of that. So One of the great things about GIS is there's an integrative technology. It can pull data from all kinds of different sources and overlay them in a spatial context. So you can look at your, you know your land use in your zoning, you can look at your crimes, you can look at the infrastructure, you can look at lots of different things on top of each other and identify New things that you never really knew about before. So right now, yeah, it's critical to solving the the complex problems that these agencies are dealing with.

Marc Culver:

You know, before we transition into AI, I wanted just to kind of have a little bit of a conversation and you know I'm a user of GIS and you know I was a public works director for the last eight years and so I had staff that, would you know, use GIS to get me data. I mean, I I used to be a big, you know, user of Arc Reader and that to get stuff, and at some point I stopped even doing that. So you know I might not be up and up on on everything, but when did, when did Esri and Arc GIS transition from a, an app base or even a local server base type of Solution or model to that cloud based model and what does that mean for the user, like what are the advantages of that it is? Maybe talk a little bit about that, adam.

Adam Carnow:

Sure. So you know we had been in the business for over 50 years and so you know when we started, it was, you know, mainframe, and so as a technology company, we have to follow the trend, and so we are always. You know, our R&D folks are always plugged into the bleeding edge of what's happening, so we're never surprised about what emerges. So, as things changed in all things, technology base, that's when we change. So you know when you know, used to be you know, individual like John and I learned on the Unix workstations and everything on a local drive, and then it went client server and then it went, you know, as soon as the internet and the cloud became a thing, that's when we moved to it. We had to because everybody was moving to it, and the really amazing thing that that did a lot of that is that really opened up the ability of real time information. Right, so it used to be. You wanted to collect information in the field. You had some sort of device. You went out into the field, you collected it all. Then you came back. You actually physically plugged it into a computer, you downloaded it and then you looked at it. Right. So now with cellular networks in the cloud. You can have people collecting information and see it, you know, near real time in a dashboard on a screen back in the office so I can see where my step, my field staff are and see the information they're collecting as it's happening. You know, you can have live feeds from traffic cameras and I can see an accident happen right in front of me and then alert people or whatnot. So that real-time capability and then, as you know, these devices you know, took off. Everybody becomes the gis user. He was holding up a cell phone for those. Yeah, I'm sorry, sorry, but yeah, yeah, so everybody became users and so, and then the other thing that really helped with the cloud is the ability to make the gis much more ironclad and enterprise ready. Right, I mean, so it used to be. You had to have all your computers in a farm and you know, stand up your app and then, like you, would stand up a voting app on election day and then you know everybody hits that app and you could bring down your whole system. Well, now you can set that up in the cloud and it will automatically Spend up more servers if needed or whatever so apps don't go down anymore, and certainly with storms and you know Disaster recovery and all that kind of bulletproofing of it. And then the gis has really cemented gis as an enterprise level mission critical system because it can ride all of that same infrastructure.

Marc Culver:

Yeah, I think that's an all episode on that. Yeah right, absolutely, absolutely, and I think that's a great example and comes back to, like you said, that mission critical aspect of gis.

Adam Carnow:

And it also made it much more affordable and quicker and easier to use. Right, you know, create an application, a sign management application, from scratch, and you know, hammer it out with some code and you know, ended up on your own machine and now it's literally check a box that's deployed and you go.

John Shain:

So yeah, it's just revolutionary. Well, the nice thing about it, too, is it fits a lot of different molds. So there's still ways and there's still reasons that you might want to deploy ArcGIS Enterprise on-prem, or maybe not even on-prem. Maybe you want to deploy it to your own cloud environment, right. So there's a lot of different ways that it can be implemented to your organization. Arcgis Online as a I guess it would be considered a SaaS. Right is still software as a service. Software as a service is a fantastic solution to get people up and running literally in 24 hours Right. To have something up and running and to be able to deploy applications almost immediately is amazing.

Adam Carnow:

Yeah, and it makes it flexible and it's scalable and much more affordable to a wide range. But we're always big on you know a lot, addressing the market in as many ways as possible. So we don't want to tell people, oh, we're now a SaaS company, you have to put it in the cloud right now. We will want to make sure that people have the flexibility of, you know, putting it on a local box, putting it in their own server farm or putting it on the cloud. So Esri's always taking that notion of make it as flexible or scalable as possible. Don't force people to in a direction they don't necessarily want to build Cool.

Marc Culver:

Aren't ready for it. So talking about you know? Interesting segue. Maybe you're not ready for this, you know, but get ready, ask in your seatbelts, because if you're not using AI now, you will be. So let's talk about that. Let's dive into artificial intelligence and what that is and how people are using that in the ArcGIS world. Hey everyone, I just want to take a quick moment to thank our sponsor, bolton Mink, who is producing and editing our podcast.

Bolton & Menk:

At Bolton Mink, we believe all people should live in a safe, sustainable and beautiful community. We promise every client two things we'll work hard for you and we'll do a good job. We take a personal interest in the work being done around us and, at the end of the day, we're real people offering real solutions.

Marc Culver:

And Adam, why don't you start us off by maybe just talking a little bit about the difference between, say, like just a simple algorithm or a filter or something, and and AI? You know what is that difference? Where is AI going beyond just some, maybe basic analysis tools?

Adam Carnow:

Sure. So the first thing is that people should understand that AI has been around for a very long time. It's been around for decades, like since the 50s and 60s, as you know concepts, especially in the academic world. So AI is not something new, but it's becoming much more prevalent and accessible due to the advancement in technology, just like everything we've been talking about just like GIS, right Computers can handle it now at a much more accessible level to the general public. Right, so it's an echo of what happened with GIS that we just talked about, right, so it has been around for a long time. And then the other thing is that people think of AI. You know a lot of me growing up and watching, like you know, the Terminator movies with Skynet, you know, oh no, you know AI. Right, we need to watch out for AI. But AI is this wide spectrum of things and it shouldn't all be lumped into certain areas about. We need to be afraid of it, or where you know danger or what have you, and so it's actually been in our. We had pieces of AI in it for decades, so you can Interpolate an elevation surface based on certain elevation points or based on contours. That is AI learning, and actually you know estimating the elevation between the area. You could do land cover. You know mapping based on aerial imagery so you could teach the computer to recognize this is a forest, this is a road, this is a building Right. So we've had that in there for decades as well. So it's not new to GIS or to Esri, but some of the newer things that are emerging from it that people are hearing about it, seeing more of, are much more applicable and are changing things. And so, while AI is the you know the broad spectrum of things, when you start really getting down into it and talking about it with GIS and Public Works, you know you need to talk about the segment known as machine learning, where you can actually teach a machine to do something and then you do that through deep learning models. So there's AI is the big circle, machine learning is within that and then within means, machine learning or deep learning model. So if people want to get really nerdy, we can, we can talk in that area. But so filter algorithms are really just math, right, I mean, you can have a filter, you know this, you know this field equals whatever. This. You know definition. Show me those are, you know. Or I can have an algorithm that estimates the elevation based on these things. That's really just advanced. Facial math or statistics, machine learning, really. Or AI really gets into doing much higher level things like visual perception, like looking at a photograph and saying, is there graffiti on this wall? Right, being able to recognize graffiti in a photo, right, that's very advanced, right. Or speech recognition right, you can talk to Siri or Alexa or Google and they understand what you're saying. Right, that's speech recognition. Another one is decision making. It can say you know if, if this measurement goes above a certain thing, I need to send an alert to somebody, right? You've got like a SCADA system and the pressure loss Okay, send an email to so and so because there was a problem here, right. And then translation right, you can speak English into the computer and it give you back Spanish, right. So that's kind of examples of AI that we all should be familiar with because we deal with it in our daily lives, right, and so that's kind of the separation, how I see it, at least.

Marc Culver:

So, and just to kind of clarify maybe I'm a little off base but like, as you were talking about it and talking about machine learning, like every AI application has had some sort of machine learning component to it, like there's been some sort of machine learning element in order to get that, develop that artificial intelligence right. Yes, exactly yeah. Yeah.

Adam Carnow:

And now, when you start talking about integration of AI with GIS, we use the term, and I think the entire community uses this term we call it GEO AI. So when you're talking about spatial AI or using your GIS integrated with AI, or AI within your GIS, the term should be GEO AI, and that's where you're dealing with a spatial component. Yeah, Okay.

Marc Culver:

Alright. So, and you know, I had the opportunity to meet Adam and San Diego at the recent PWX conference there and we talked a little bit about this episode and that and I think we had a 10 minute conversation like we could have recorded part of this podcast right there on the floor, yeah, but you started to talk about some of the you know applications and solutions that are being used right now that are using AI and you touched on them really briefly. But why don't you go into a couple of them, and in a little more detail right now? And I think one of the more interesting ones was the, the, the PED ramp identification one. Yeah, I think that was in. Was that New England or?

Adam Carnow:

Nebraska. No, it was Douglas County, nebraska, nebraska. Yeah, so we're Omaha. Yeah. So when you talk so we talked about AI and all that. Now we talked about GEO AI. So one of you know there's three major ways that GEO AI provides the value. The first one is object detection, so looking at objects from imagery, from videos, from point clouds, and so examples, or like land cover or change detection, right, a lot of times, like property appraisers, they can fly aerial and then you fly another year and all of a sudden, a pool appears in this person's backyard. But they didn't, you know, know it. So we need to up the property value because they have a pool now, right. So that kind of thing. But there's lots of things you can detect out of imagery. You know signs, fire hydrants, traffic lights, trees. You can have cameras. You know who the Hillsborough, oregon. They're putting cameras on their street sweepers, so they're, you know, they're sweeping the streets every day. They're collecting pavement condition information. So you can now automatically have a AI routine or a model, look at this imagery and pull out cracks or potholes or whatnot. So that's a big piece of the value here is this object detection, especially from imagery, video and point clouds, and so the example from Douglas County was really interesting and the story I love behind it is that you know, it was the GIS analysts that kind of changed things right, because they knew that they had to inventory all their ADA curb ramps across the county because they needed to know how many they had and, you know, start the asset management lifecycle with that. So what were they going to do? Right, they were going to send crews out into the field to collect them, and they had collected some as they were put in as part of projects over the years, and they had about 16,000 in their inventory and they weren't sure how many they had. But it was the you know. So they turned to the GIS analysts and said you know, you need to set up an app for these guys to go out into the field and collect these curb ramps. And he was the one that said wait a minute, let's try to think differently, let's not use this muscle memory and do what we always do and send crews out in the field. We've got one inch you know resolution imagery for the whole county, which is really good imagery. They pay a lot of money for that, right? I mean, you need to extract as much value out of that as you can. So he said I've heard about this AI thing. Let me see if I can get the computer to figure out if it can see ADA curb ramp in the imagery. And so he worked with somebody at Esri and started with one of our object detection models and they started training it and, sure enough, part of it is just like teaching a child. You have to get them to understand what a curb ramp is. So it's proximate size, proximate shape, a proximate color, et cetera. And so, like this is one of the interesting things that comes out is what's about the same size and shape as an ADA curb ramp but a sunroof on a car? So it started pulling out all the sunroofs on the cars, right? So it's like, okay, that's wrong. So now we have to figure out how to teach it not to do that. And then the other thing is you know it doesn't have to look at every square inch of the county. We know they're at intersection, so we can take only the imagery at the intersections and just analyze that for portions of parking lots or whatnot. So you get more intelligent as you start working down this road and then you know some of it. You can't see it all because there's shadows, right, or maybe things in the way, so but anyway it turned out that very quickly he was able to come to a model and no AI is perfect Okay, no AI is 100% perfect, so that's another thing to realize. There's always going to be some sort of human interaction and there also needs to be some sort of data quality check. But he was able to get to know, you know, 95% plus identification of these across the county, and turned out they had 34,000 of. You know they started with 16,000, had no idea how many they had, and he was able to do this on an, on an individual PC running our GIS pro and eventual model. When it ran and identified the 34,000 plus it was, it took about 12 hours We'll think about, you know they did the math on how many hours it would have taken in the field for the crew to, and it's saving the month, month of staff time. So really great use case that applies everywhere and I love it because it's getting more value out of that imagery we should pay a lot for. So yeah, Great.

Marc Culver:

What would you say is the average resolution that agencies are using on their areas Like?

Adam Carnow:

Oh jeez, I don't know. I mean, it depends, john, you got any.

John Shain:

Yeah, I mean I typically in the Midwest. Now that's been constantly improving as well, and we're seeing a lot of data sets delivered now that are two to four inch quality are pretty typical, and those are even in rural areas right. So we're talking about fantastic resolution now. That's great.

Adam Carnow:

Yeah, that's great, yeah, and so you know that is expensive data, so you need to look at ways to extract as much value out of it, and so AI is a great way. What can I pull out of that? Whether it's building footprints, whether it's pools, whether it's blue tarps after a storm comes through, so who's rooster damaged? Yeah, there's all kinds of things you can pull out of aerial imagery Again pavement markings and cracks, fire hydrants, traffic lights I mean, you name it.

Marc Culver:

Well, and that gets into merging a previous topic of ours and that's drones. You know with that as well and you touched on like a post storm survey and going through and just taking drone footage of as much of the areas you can and then do all sorts of things with that video. Or you know that aerial imagery, you know as far as the tarps that you're saying, or trees down, or roads blocked, or standing water or what have you, and you know maybe I would imagine that there that there will be AI tools to help you with those as well. So, yeah, for sure.

Adam Carnow:

Another great one on the public work side is impervious surface for stormwater billing, yeah, and stormwater modeling right, and so used to be for a lot of agencies. You know they would fly their aerials once a year. So they would fly the aerials, right, and then they calculate the impervious surface for all the parcels, send people their bills and then they'd wait another year, right. And then they do it again for all the new places. But what if the day after you fly one, a new thing is built? That means those people aren't getting billed for a whole year, right? Well, now, as soon as they get their certificate of occupancy, you can run out there with a drone. I can fly the property in 20 minutes, come back, start billing them from day one, and I've had agencies pay for their entire drone program with that. What application. Wow, wow. Yeah, so it's really, there's a lot of area for creativity and innovation.

John Shain:

That's a fantastic example because that's exactly what's happening. You know, we might be working on cycles of one in a in the best case scenario, sometimes three year sorts of imagery in order to reestablish in previous areas and to be able to deploy a drone within a day to be able to get that information out and append it to the source data set in order to initiate the billing. I mean, that's just, that's a fantastic example.

Adam Carnow:

Yeah, yeah, that's another note you get. Oh, I guess you said you did do a drone episode. I was gonna say you had a whole episode on drones.

Marc Culver:

We did. I mean, that's not to say we won't do another one, talk about maybe some more specific applications, but we did do one of you. So if you haven't heard that yet, if you're listening, you haven't heard the drone. Listen to the drone episode. Yeah, go back. I don't remember what number it was. I think it was like eight or nine or 10, but right in that area there.

John Shain:

You know, one of the things I you know, back to what what Adam was talking about before with building and training this particular model for identifying pedestrian ramps, those, those particular models, those image libraries all of those characteristics can also be shared and reused, right, and that's an important component too. So sometimes thinking about building a model from scratch is can be, can be a daunting exercise, right. So the more information, the more models are trained, the more information that becomes available out there for people to be able to leverage. Those is amazing, and I know that not too long ago we were using, we found, a library to be able to identify passenger vehicles versus freight vehicles so that we were able to run analysis against captured video to be able to do traffic counts, right. Yeah, that's great, and classification. Yeah, as well. So you know how many freight vehicles are coming through this intersection versus passenger and so on it. And as those libraries get better and better, they're going to be easier to access and to plug in against some of those. You know foundational data sets like aerial imagery.

Adam Carnow:

Yeah, that's a great point, and we have a collection of models that are developed ready for you to start with, so you don't have to start from scratch. Yeah. And then we encourage our user community to share theirs as well, right.

Marc Culver:

So you know, just kind of on the flip side of all the benefits of AI and I think you touched on this a little bit, adam but what should we be weary of when we're using AI or someone says that they're using AI to complete a task? What should we, what questions should we ask, or what should we kind of be checking on?

Adam Carnow:

Well, first of all and I just listened to an interview and I should have sent GEO AI Sorry Well, that's okay. I just looked to a really good interview yesterday on AI and it came from the standpoint of Tempe, arizona. They just released. They're one of the first municipalities to release an AI ethics policy. Oh. So they've been using AI. Tempe is very innovative and they were interviewing their chief data officer. She was talking about that policy and it's got a lot of press since it came out, and I really like one of the concepts she talked about. She talked about how people need to understand that AI is not it should not be used in where you tell the AI go do this and buy your independent and all your own. You know it's a collaboration between a human and the AI, so there is no way that ever you should just say here, the AI can just do this on its own, of its own, with no balance. Check the balances. No, no, no. It's a collaboration between human beings. So the first thing I would say is you know what's that level of human involvement? Is there enough to counterbalance any of the dangers that AI can bring to the equation? Right, because models are written by humans, so there is unintended bias, just the way our brains work in there. You need to make sure there's that. How are you reviewing the results? What are they doing with the results? You know what's that level of human involvement? Do you have enough of these checks and balances in place so that it's bringing the value, but none of the risk or limiting the risk as much as possible. You know there's lots of opportunities for AI to bring value, but we have to make sure that, again, it's a collaboration and it's not. We don't further any bias that's already built into systems or processes and you know so. Here's another great example. So we talked about aerial imagery and extracting the curb ramps and so forth. There's a whole nother area of imagery where you're using what we you know with no terrestrial imagery, so you can put camera on the side of a truck and take photos as that truck is driving and then use that to identify things and actually, you know, triangulate and you can locate things. So, like I'm here in Charlotte, charlotte, north Carolina Water Department, they were spending $300,000 to $400,000 a year to survey companies to locate all of their water meters. Well, I hate to tell the survey industry, but they changed that. They put cameras on all the city vehicles that drive the streets every day and they take pictures every so many feet and that is then spooled up to the cloud where it's analyzed by a machine learning model and they can identify and give you the location, within a pretty good accuracy, of all the water meters in town and they're saving that $300,000 to $400,000 a year. 10p is doing it with graffiti, so they had a bad problem with graffiti. They put these cameras on the side of their trucks the trucks are driving, takes these photos. Once it analyzes the photo, once it finds one that has graffiti in it, it automatically fires off a work order to remove it. So the minute that tagger is done, spray paint in their graffiti. They know about it within a day or two and they remove it. And it's revolutionized their fight against that.

Marc Culver:

Yeah, and before I asked John where maybe we've done some AI stuff, and I know one example where we're thinking of one or where we've been talking about it. But I just wanted to touch on the water meter thing because this always gets me Like whenever I go to a place like San Diego or North Carolina or the Carolinas or Florida or something, and I'm walking the streets and I see this little cover and it says water meter on it, and I just shake my head Like God, how much easier would our life be if we could put our meters, our water meters, in the public right away, so we didn't have to get into people's homes every time we needed to swap a meter out or a radio goes bad on it or something like that. And so you know, for all of my fellow public works peers and our nerds in the cold weather climates, we just kind of shake our head like God. That would be so nice, but anyway, just had to share that moment of awe there for the warm weather climates. But yeah, what have we done? Like, where have you seen? Maybe some of that?

John Shain:

Yeah, well, I think I would echo some of the comments that that Adam had before, where we've been using things that are now termed AI that were essentially automation and running models. So we're very fortunate here to have very strong Python developers who've been able to aid us with automation and automation of combing through large datasets, but also doing things like being able to interpret in previous areas and to be able to use, you know, object classification to be able to determine again where vehicles might be or open parking spaces or you know things of that particular nature. One of the things that I find really interesting right now is the is the use or the application of AI on top of of IoT in real time information. So we've talked about this quite a few times, about what happens if we have real time data coming in based off of flow, or how infrastructure should react to, maybe, a weather event that's upcoming, and what are current models, and being able to use AI to be able to adjust things in real time. And, of course, all of those things are done with human supervision, you know. I think we all need to understand that at this particular time, but to be able to, to be able to implement that sort of intelligence into our system is something that I still think is going to be controlled mainly from a geospatial or a geo AI type of a system.

Adam Carnow:

Yeah, yeah, that's a great. That's a great point, john, and I mean I had notes in here. You know, if I had a whiteboard, I would do it on here. But one of the main drivers for the need for AI is real time data and IoT sensors. And so the flow starts with. You know, organizations have tons of data and they've been spending a lot of time and money to manage it and collect it over the years. They're finally realizing that maybe they should do something with that data, right? And so organizations are trying to become data driven, right, they're trying to make data driven decisions. So if you then accept that this organization is going to be data driven, then I need, I want the best data that I can get to make that decision as best as I can, which means, okay, real time data. So what are they doing? They're deploying IoT sensors, internet of Things, whether it be, you know, road temperatures or flow rate or flood levels or anything would have ever did. So these things are streaming data 24, seven at some interval. Well, that leads to big data, and that leads to big spatial data, because one thing that every sensor has in common is a location. It exists somewhere, either stationary, like a flow gauge, or it moves around like a vehicle, right, and so these things have a location. So you got to use GIS to get the value out of that data, and that leads to big spatial data. Well, that data gets so big very fast that the only that humans can't really effectively analyze it, and so you do set up the automated routines where an AI can start to monitor this and then, when it sees something going, it can then alert somebody. So that whole workflow and evolution of our agency is leading to the need for more and more AI, to get bigger and bigger because of the real time need for the from the data driven decision.

Marc Culver:

Yeah, yeah, absolutely, and I yeah, it's like you said we're we are collecting mammoth amount of data in so many ways, but how can we get better at analyzing and doing stuff with that data? You know, as we wrap up into the last five minutes here of the of the podcast, I wanted to ask both of you, like, where do you see GIS in the next 10 years? Like, what will be the next big innovation? And I think you know AI is going to be a part of that. You know, like we've been talking about for the last 25 minutes, but you know what, what are the next big things with, with GIS, like, and maybe even on a stretch, just like I kind of maybe see something here, you know, but where do you think maybe some big innovations there? John, you want to?

John Shain:

go first or I'll let you go first on this one, okay.

Adam Carnow:

Well, as we do more and more real time, what we just talked about is going to become even bigger, but the whole, one of the areas that we haven't really talked about yet that, I think, is where the future of AI is in prediction. So there's one thing with looking at a data flow and alerting someone to when something happens, or being able to analyze the data or, like you know aerials and pull out your curb ramp, but there's a whole new level of intelligence when I can predict, based on historic information, so I can predict on where and when there are going to be traffic accidents, right, I mean, that becomes incredibly important as far as you know saving life and you know property, et cetera, and so I think the prediction realm is really where you're going to see. The growth is. As these models get more and more sophisticated, they'll be able to predict things, and now you start thinking about minority report, right, and then in the movie and predicting the crime and all that.

Marc Culver:

But that is the second time in two days that I have heard of reference or made a reference to my order report. That's great.

Adam Carnow:

That's awesome, you go. Yeah, it's time. So you know, one of the ways I can example is, I was talking to a customer recently they actually have this model that they this was in Florida that the Army Corps developed a long time ago and they wanted updated, and it can predict debris from debris volumes and types from a like a storm before it hit. So if I live on the coast of Florida, I'm a county or a city and I know that a cat for hurricane is going to hit this area, because I have the track, I can run this model and it'll, you know, estimate based on it, knows what exists in that area, what kind of buildings and land cover, et cetera. You're going to have cubic yards of concrete and cubic yards of plywood and cubic yards of downed trees or whatever, so that I can then call contractors and get debris haulers ready in the area to go the first minute that storm is out of there. So think about how that can you think about resilience? You think about sustainability, the climate change. I mean all of that that that's going to help us deal with the situations that we're dealing with and that, to me, that's the value of that. You can't even explain it, right. I mean you can't even put a dollar on it when you're especially when you're talking about disaster recovery or resilience around that kind of thing.

Marc Culver:

Yeah, and I think another really great example of that and it's very related to what you're talking about but you know, like anywhere, you know, even in the non-hurricane prone areas, in Roseville, where I was a public works director, we had an area underneath an interchange, a low spot that would flood at a certain point, given a certain rain intensity. And again, I think using a tool that can even monitor some of the rainfall that was happening at, you know, as it was happening, could send an alert saying hey, you're going to have flooding at this intersection in 20 minutes Yep, you know some of that predictive stuff and extend that to other parts of the city. I think that's huge because you can go send resources there to deal with that before cars are driving through it and get stuck or people are drowning somewhere, yeah Well great John. Did you want to?

John Shain:

yeah, I wanted to say, you know, as a GIS professional in the upper Midwest, where I see the next 10 years is there's going to be I believe it's going to be a lot more accessible to get real time information. Yeah, more sensors are going to be able to become available and GIS is going to continue to integrate with all of these different platforms that spool off and spawn information to be able to consume that. Now, they've had tools around for some time, but I think that those things will become more easily accessible where we'll be able to start tracking certain measures in real time, much easier than what we're able to right now and much less expensive as well.

Adam Carnow:

Yeah, yeah, I think it's going to. You know it's going to. You're going to start getting I'm going back to asset management. I mean you're going to start to get predictions of failures of infrastructure before it happens. Right, you're going to know. I hope so, right, right, hopefully you know before this potholes happens, before this waterline breaks or what not, because the more and more data historical data you've got, the better your predictions are going to get. The more time, as the technology increases and the data gets more accurate and more frequent, you're going to get better, better tuned models and you're going to get better predictions. So yeah, the outlook, I think, is very good for the technology to bring even more about.

Marc Culver:

Awesome. Well, this has been great. And then love the conversation about AI and GeoAI, and I will never call it ARCMAP again.

John Shain:

It's still supported for another year.

Marc Culver:

Yeah, yeah, before we go to add was there anything else? You wanted to add.

Adam Carnow:

Well, I mean the one last piece of recommendation I would have for people. If you're not using AI right now, start immediately like the minute we get off so you get off this podcast start looking for ways that you can utilize it, because lots of agencies across the country are using it and have been using it for quite some time, and they're getting lots of value out of it, and there are ways to jumpstart your path down the use of it.

John Shain:

Awesome, John. Well, I just want to thank Adam for joining us and being part of a great organization that's really pushing tech forward right. So without Esri, this conversation might be a lot different.

Marc Culver:

Yeah yeah, before the podcast started, I think we had a really great conversation just about how you know you and stop me at any point, adam but you know we're in this environment and Esri kind of exists an environment that doesn't have a lot of competition and a lot of times that can actually lead to kind of a stagnant environment and technology If you don't have a competitor, that's pushing you. But I think I got to give Esri credit for really pushing themselves and allowing the users to push you. So thank you for that and good job.

Adam Carnow:

Well, you know, as Jack Danger and our president and founder always says, we work for our users. We don't have shareholders Our users. What they, we listen to them very intently and whatever they need, that's what we try to deliver. So that's the whole theory behind this company that's taken us through five plus decades.

Marc Culver:

So yeah Well, thank you for taking public works on that journey with you and giving us a really, really amazing and transformational tool. So thanks again for joining us, adam. Thank you, john, for your time today. And one last thing before we go if you have enjoyed this episode and the podcast in general, we ask that you help us spread the word. If you're on link to end, please like and, more importantly, comment on this podcast. And you know we're also on YouTube, so you know you can watch us on YouTube, see all of our pretty faces on video here. You know when John flashes a cell phone you can see that. But hey, help us out, retweet one of our posts reposted or, better yet, tell your colleagues about the podcast. Thank you, nerds out.

John Shain:

Thanks to us.

GIS in Public Works
Advantages of Cloud-Based GIS Solutions
AI Applications in GIS
Benefits of AI in Geospatial Analysis
AI in GIS and Real-Time Data
Predictive Modeling in Disaster Preparedness
Company Focus