HxGN RadioPodcast

Public Safety Now: See the unseen: The value of AI in public safety

Do you have an operational blind spot? Artificial intelligence can work behind the scenes to support emergency communications professionals and identify issues that could not be detected by human observation alone. Learn more about the role of AI in public safety in this episode of Public Safety Now.

JWh: Hi, and thanks for tuning into Public Safety Now on HxGN Radio. I’m your host, John Whitehead, Vice President of Sales for U.S. Public Safety here at Hexagon’s Safety and Infrastructure division. You know, over the last few years, public safety has really become complex – not that it wasn’t before, but it’s just the amount of calls, the amount of incidents that are coming into an agency. The data that’s coming in is just overwhelming, is how I’d put it. Today, we’re going to be discussing artificial intelligence, or better known as A.I., within public safety. We’ve got Jack Williams, our Strategic Product Manager, here from Hexagon to discuss how do we implement A.I. and how can A.I. kind of help out with public safety and help us kind of filter and utilise all of that data that’s coming in to be better in our job. So, Jack, welcome. How are you doing?

JWi: I’m doing good, John. I’m ready to talk some A.I. and how that can apply to public safety. So, thanks for having me.

JWh: Yeah, it’s interesting because, you know, there’s always been the backroom joke, right? The movie like Minority Report where there’s all this intelligence, and it’s foreseeing the future, and it’s telling us all these things. And, you know, it always feels very Hollywood. In real life, we’re answering 9-1-1 calls. We’re responding to incidents. We’re catching the bad guy or putting out the fires or helping the sick. Those Hollywood things always seem to be just that, just in the movies and not real. But, man, once we start talking to artificial intelligence, we really start bringing that intelligence into public safety. How do you see that helping us?

JWi: Yeah. I mean, so anytime you hear that word artificial intelligence, you know, you start to think of the Hollywood movies, right, John? And all the different I, Robot and Terminator and Minority Report and all those futuristic bad scenarios.

JWh: Yep.

JWi: That’s not what we’re here to talk about. And in reality, A.I. is still at the point where it’s not quite mainstream yet, but it’s becoming more and more. I’d say within the next five years it’ll be implemented. It’s already implemented pretty much and maybe don’t even know it. But yeah, it’s nothing to be too scared of at this point. I don’t think we’re hitting the singularity quite yet.

JWh: Yeah.

JWi: I’d say when it comes to public safety, being able to what we call operationalise A.I. so that it can be assistive for the human is really the major benefit and where I see A.I. machine learning, really helping public safety agencies be more efficient, make better decisions. So, I’m excited about it. So, don’t get too scared yet, John. We got a few more years.

JWh: All right. All right. All right. Well, you’ve talked me off the ledge on that one, so that’ll be good. It’s funny, and I really think it is exciting. I jokingly say I’ve got artificial intelligence here in my house, right? I got this little machine, and all of a sudden it wakes up at a certain time and it goes around the floors and it vacuums up any crumbs or dog hair or anything else that might be on the floor. And it goes back home, and it empties itself. It’s just a cool thing as we’ve started getting into this world. And I looked at that as, man, all these things are new, and all these things are so cutting edge. But I did a little bit of research, and artificial intelligence isn’t new. This stuff’s been around. I mean, I’m looking here. It says, in the mid ‘50s, John McCarthy was credited as being the father of A.I. So artificial intelligence isn’t anything new. This stuff’s been around for a long time. And I think it’s really kind of cool as we talk about public safety and how A.I. plays into that. I think it’s just exciting to kind of see us step into that realm and to better utilise, again, as I said at the beginning, some of that historical data and how we can kind of use that to better our responses.

JWi: Yeah. I think A.I. machine learning really has the ability to help us see the totality of our data, right? And so what we lack right now is sort of the ability to process large volumes of data coming in in real time, and A.I. machine learning can be used to filter the noise, if you will, and provide proactive insights. So, that’s one thing. But back to your point about A.I. Yeah, A.I.’s been around for a long time. I mean, A.I., it’s a broad concept of basically having a machine being able to carry out tasks in a way that a human would—considerate, intelligent, or smart. And so that could be something such as a rule of thumb or heuristics or a decision tree, or you can get into deep neural networks, and there’s all different kinds of branches of A.I. and approaches of machine learning. And just to kind of clarify too. You might say what the heck is machine learning relative to A.I.? Well, machine learning is a subset of A.I. that’s specifically focused on you set up an algorithm, train a model, and then that model learns over time by itself from experience without having to be sort of programmed, right?

JWh: Right.

JWi: And so, there’s a lot of potential there for public safety to leverage more of their data and also have more data at their fingertips – relevant data at their fingertips – in real time. And I think that’s where A.I. machine learning can really help.

JWh: Yeah. That’s interesting. Now, as I said, thoughts go through your mind as we start talking about this stuff, and I’ll just be open and say I’m assuming that some of the listeners of this podcast are probably thinking, oh, here we go. This is where it starts. Next thing you know I’m not going to be needed anymore. There’re not going to be a need for dispatchers or we’re not going to have to have my position in here. They’re going to try to automate all of this stuff. How do you think this is going to affect—we’ll just start at the dispatcher level—how do you think it’s going to affect the dispatcher’s job, and how do you see the A.I. fitting into that space?

JWi: The way I see A.I. fitting in is not as a tool to replace the human, the user, the dispatcher, the call taker, the supervisor. It’s assistive. And that’s an ethical and practical way, especially in public safety, of all spaces, to leverage A.I. As bold proclamations that you’ll hear Elon Musk and some of these guys say truly having A.I. and robots and software being able to carry out and have that human judgement, that human emotion, I mean, that’s still quite a far ways away. And I don’t know if we’ll ever get there. Don’t want to say never, but it’s a long ways away. So there’s an area where the real, practical, operational way to leverage A.I. for public safety would be to help augment, meaning to add to a person’s capabilities, giving them more information, helping them when they’re stressed out, helping them when they have information overload, which is very common. So the dispatcher / call-taker, they can benefit from A.I. by simply having A.I. running and constantly scanning CAD and dispatch data sources, looking for connections, looking at the big picture, trying to find trends so that you can relate things in real time as opposed to after the fact when it’s too late. And so, at that operational level, from a dispatcher point of view, A.I. should never automate or replace a dispatcher / call-taker. That human judgement factor has got to be in there. But what A.I. can do is really almost make them like a superhero, right?

JWh: Yeah.

JWi: You know, we’re trying to augment. So, I think it’s got that capability to enhance and assist.

JWh: Well, and it’s kind of how we started off this conversation. There is a lot of information going on in all of our agencies. It used to be where, as a dispatcher, a police officer, you’d worry more about my policy and procedure in this type of incident. I need to do a, b, and c. That’s what was going through your mind. I think that’s still the case today. But I think there is an influx of calls. I think that for the most part, we’re seeing more and more calls for service coming in the door and people asking for assistance. And I think what you’re telling me is, and this is what it sounds like, is this is really just an aid. It’s another tool in the toolbox, if you will. Because information occurred in the past, but yet we’re focusing here in the moment. This is going to go ahead and be able to do some of that research for you and provide you with actionable data, if you will. You know, I’m handling the incident that’s sitting here right in front of me. There’s a lot of other great historical data that might be able to assist my responding units. So this product is going to be able to go look for that, do the research for us, and then prompt us and give us information that we can share with the responding units to help them better do their jobs.

JWi: Yeah, John, it is. It’s a tool that has a lot of potential. It can help with that information overload where today how that’s done is people use different tools inside of the agency, so in front of their screens. So, they might go do a search here. They might go look in CAD and drill down into a specific tab and look for supplemental information. They might go Google something. All those things are done sort of manually today that a person has to kind of go in and do what we call contextual search. The other way you could have this ability to handle these high-pressure situations in large data volumes is just simply through pure experience, and that just, obviously, takes time. So, where it can help is really filtering out—and it is a tool. It’s something that doesn’t exist today. So, I don’t think it’s something that should be intimidating, by any means. It’s an asset that you can leverage that is more than what’s there today. And that’s very promising. And it’s also nice to know that the system itself is smart. It’s going to constantly sort of aggregate and find patterns and similarities and anomalies and let you know of them. Now, it’s not going to take any action. That’s the human-decision element. But it has a lot of potential, just even for front line, but also other roles in the agency who might not use—they don’t actively dispatch or take calls.

JWh: Yeah. And it’s exciting because Hexagon released this product September here of 2020, this Smart Advisor product. I’m excited that it’s bringing that A.I. aspect into public safety. It’s definitely the first of its kind, I think, in the industry. And I think that it really is going to assist. I keep going through different scenarios in my head, right? You’ve got a person—I’ll call him Joe Dirtbag, right? He’s living out there at this location. He has a history when it comes to public safety. Maybe he assaults officers. Maybe he’s just known to carry weapons. As responding units, they may not have all of that history. We may have a new officer. We may have someone out in the field that’s not familiar with maybe that region. They’re working a different beat that day. I love that we’re going to fill in the blanks, because in my mind, this immediately tells me, you know, this Smart Advisor, or this A.I., it’s going to save lives in the long run because it’s going to let that officer know, hey, be on guard when you approach this person, or heads up. This is what’s happened in the past. And that’s really what it comes down to, is just saving officers’ lives.

JWi: Yeah. Saving officers’ lives and saving citizens’ lives, too, by getting ahead of things before they escalate out of control. Yeah, I mean, that ability to draw back on information that happened in the past and compare that intelligently against what’s happening in the here and now, that’s very valuable. You know, we do that all the time in our heads. But, you know, I don’t know about you, John. The older I get, the harder that gets.

JWh: Yep. Yeah, that’s for sure. And you know, what’s cool about this is that in the dispatch centre, we’ve all got that one dispatcher. All the agencies have that one dispatcher that everybody looks to. Maybe it’s one dispatcher on each shift, but they’ve been there forever. They know all the alley ways. They know all the shortcuts. They know all the frequent fliers that are calling in to 911. And I’ve talked to some of my friends in the industry and they’re saying, you know, when those people leave, that’s a lot of knowledge that leaves with them. That knowledge isn’t anywhere written down. It’s not in any book. And now a new dispatcher comes in. And I remember back when I started as a new dispatcher, there’s almost this overwhelming feeling. You’ve got to learn all the policies and you got to learn the procedures. What’s this computer in front of me do? And you’ve got to do all of this different training. And then in the background, you’re also concerned because, wow, I’m just not as knowledgeable with the past or with the historical data that so-and-so is sitting across the room. Historically, dispatchers turn around and ask that person, “Hey, Susie, you know this guy, Mr. Smith, that lives down here?” “Oh, yeah. Mr. Smith’s lived there for 30 years. That guy doesn’t like us, man. Just be careful. Send two units out there.” “Oh, okay. Great.” I think that that’s the other level of excitement that I’m seeing with this product as this can assist our new dispatchers. We can bring new dispatchers, get them ramped up quicker because they’re not going to worry about memorising all of those types of things. They’re going to be able to learn on their own time, which is great. It’s always good to have that inherent knowledge. But because we’ve got A.I. running now in the background, we can supplement them and give them the information that they need so they can focus on the real work that I need them to focus on.

JWi: You hit on a very key topic there, John, and I think there’s a couple elements to it. One is there’s a major issue of PSAP or call centre turnover right now. And it’s also challenging to find new employees willing to take on the stress. I mean, it’s no surprise that the turnover rates are high, to be perfectly honest, because it’s a high-stress environment. There’s information overload. You feel like you’re, you know, you’re always on defence, right? And what A.I. can help us do is kind of pivot that and really help us— you still have to have the human, but what it can do is give you that confidence, that safety net, that something else is there sort of watching out, looking for some trends, anomalies, kind of looking at the bigger picture, making sure you don’t miss anything so that, ultimately, you make the best decision. And that in of itself reduces stress for a very stressful job already and allows new staff, new dispatchers, like you said, to have that sense of, okay, I know I have an airbag, like in a car, and I have something that’s kind of watching my back, if you will. And it will result in less stress. And I think, ultimately, when you’re not under stress, John, you’re going to learn better. You’re going to learn that experience. And then for the other aspect of it is for those people who, like you said, are those that one guy or one girl in the call centre that—

JWh: Yeah.

JWi: —knows the lay of the land. Well, they’re busy all the time sort of facilitating their insight and knowledge to people, and they’re called on for the tough tasks. So what A.I. can do is help free up their time, supporting, say, some of the newer employees, and then they can actually take it to the next level and go in and configure the A.I. and tweak the settings, because they have that intuition. They know the area. They know the region. And so that element, too, is not just like you turn A.I. on and it just does it. The really, you know, intelligent, aware, experienced people can go in and actually tweak the A.I. so it looks for specific things. So, I look at that as like it amplifies those all-stars, and it also helps the novice come up to speed quicker and overall reducing stress. It’s people-centric A.I. That was the genesis of this. Oftentimes, you think public safety, the people in the call centre oftentimes are in the background, and their job is just as important as anybody else’s. And so, what we want to do – and the whole goal of this – is to help our frontline personnel in a call centre, but also help other people in the agency, too, with real-time insights. And that was a big catalyst for why we started down this path.

JWh: Yeah. And there are some inherent tools in CAD systems of today. So for years when a dispatcher puts in an address, for example, or puts in a plate or a name and they fill out those fields within an incident, the system’s going back and said, OK, this address, we’ve been out there x number of times over the last year. It’s that historical data. Dispatchers are using that today. Officers responding to calls, they’ve been using that for years, or they should have been using that for years. This really takes it to the other level because, you know, as I go through examples and I go through some of the things that I could see, there’s a lot of information, let’s say in the remarks of a call, someone may say, someone vandalised my mailbox; he had a blue vehicle, or a blue Camaro, let’s say. Okay. I put that information in. Most times, an officer goes out, vandalism report, done and over with. But now think about how that blue Camaro, just that little phrase, could come back as that person driving in this vehicle. Maybe they go to another location. Maybe it’s a simple mailbox vandalism, but maybe it escalates. And now all of a sudden you may have a road-rage incident, or you have an escalated type of incident that happens hours, days later. That blue Camaro comment could say, hey, you know, this could be one and the same. And I use that as a simple analogy and a simple example, but I think that that’s—we’re already having historical data come back based on the address and the name and the plate. But why not have it on some of the comments in the incident? Why not have it on that other data? And in the past, I don’t have time as a dispatcher to go digging around for every call that had the word blue Camaro in it. But, man, it’d be nice if all of a sudden that alerted me and said, hey, heads up, you’ve got that here. And that’s a long version to kind of get into my next topic that says it’s not just for dispatch. This thing’s going to roll out and assist the officers responding out in the mobile as well, don’t you think?

JWi: Yeah. I mean, being able to push alerts. So, the way we approach it is we mine this data and we have these different agents that look for different things we call missions. One of the agents we have is very similar to the thing you just mentioned. It was called a similarity agent, where it will look at what I would call near-term recent events, say, last 30, 60 days. And it looks at the totality of the data. Like you said, there’s automated search features for the common fields or attributes of an event. But things like remarks, stuff like that, oftentimes overlooked because, like you said, it’s impractical to dig through all that stuff. So, being able to leverage that in real time is extremely important and that we can use that with our similarity agent to find similarities between otherwise seemingly unrelated events. So, here’s a good example, John. I think this happened in the U.K. We mocked up a scenario about this in Smart Advisor. Smart Advisor would be perfect for this. So there were some violent crime in this city, and there was a stabbing earlier in the day. And then there was a detective investigating this issue in the field, and it had the community really worried, upset, etc.. About eight hours later, another stabbing occurred – same city but different part of town. And so then two hours later, I think, there was just a random call about a suspicious person saying profanity to passersby. And in the comments, there was some slight—there was the guy was wearing a camo jacket. He was a Caucasian. He had a tattoo. But the comments don’t quite match up like that. There’s got to be some intelligence. And with Smart Advisor, we reran the scenario, would be able to detect, you would have got an instant alert that says, “Hey, we believe that this suspicious-person event that was just created is very similar to these two stabbings that occurred in the last 24 hours. You might want to—” and then the dispatcher could easily click Share and push that out to somebody in the field. Or if they have Mobile Unit or OnCall Dispatch Mobile Unit product, they could get the notifications directly themselves. So that’s a real-world scenario where Smart Advisor could be used to connect the dots. Another one is in Australia. They had this big event where there was a lot of asthma attacks occurring, and it was because of a thunderstorm churning up pollen in the air. And unfortunately, they missed it. Like, they just—you know, there was an influx of breathing-related calls and EMS calls, and they were unable to kind of put two and two together, if you will, until it was too late. And a lot of people, I think nine people died as a result of that. And they did a big study and talked about how they need to get ahead of rapidly evolving emergency situations. That’s the perfect type of thing for Smart Advisor. So, there’s a lot of different potential applications, even an analyst, John. Let’s take a crime analyst.

JWh: Yeah.

JWi: Maybe they’re investigating something. Once again, if you’re an expert, you can go into Smart Advisor, tweak the A.I., set up your own rules for the A.I. and have it say, okay, I want to monitor any time I see the word stabbing or anything related to it, alert that dispatcher to contact x, y, z. Right?

JWh: Right.

JWi: That can now happen without that—and that person might, you know, they might have had access to CAD, but it’s impractical for them to log into CAD and monitor stuff, right?

JWh: Yeah. It’s just mining that data. And it’s interesting to hear how, you know, let’s put that artificial intelligence in the background, not to take over for us, not to do anything more, but give us just actionable intelligence, just data that we can actually utilise. And you brought up some great examples. It’s things that as we’re in the day to day, we just don’t have time to go back and do that—I’m going to keep saying research—but we just don’t have time to go back and do that research while we’re in the heat of the moment. And what I like and what I’m hearing today is, is that let’s get this tool in the background that kind of keeps an eye on that data, that kind of says, hey, did you know, hey, maybe you need to know, maybe this is relevant to the incident that you’re working now. And I’m excited about that because, you know, as I go back, it really just makes us smarter in public safety to have that data in hand and really, like I keep saying, just help us overall, better do our jobs. I mean, at the end of the day, it’s really simple, right? When you narrow down public safety, what do we do? Well, we provide safety for the public. I mean, and I know I’m taking it to its lowest common denominator, but that’s really what this is. And I think as we’re out there protecting property and lives and making sure that we’re assisting our public the best, having all of that data is going to be able to help us and having all that data at hand and available when needed, it really is exciting because I think today we go out and we react. Incident happens, we respond, we take the information, and then we’ll investigate it. Okay, that’s great. I’ve got detectives, I’ve got analytics people in the back. There’s post-incident type of reporting that’s going on when you have the time to kind of go back and dig through that data. This is exciting because it sounds like a lot of that information can be brought to your attention as the incident’s occurring and not being so much reactive but actually proactive as I’m responding to that call.

JWi: Yeah. We call it a push, right? We call it a push system, but that’s a broad term. But in the tech space, I’ve heard it referred to as a push system, meaning, with the way we’re leveraging A.I. and the way it can benefit public safety, you’re right. It’s to protect the community and save lives and keep the public safe, right?

JWh: Yeah.

JWi: And so, in order to do that, we need as much information as possible that’s relevant, not just data, but really proactive, intelligent insights. And all Smart Advisor does, John, is it has all this A.I. working in the background. But you don’t know that. At the front end, I mean, you’re not seeing it. And the goal is it’s not this big, bold, flashy thing that’s going to take over your whole screen and start talking to you. No. It’s going to issue you an alert in real time, and say you might want to know that this is kind of similar to that. Or, hey, the supervisor is monitoring the situation. He says, man, we only have a typical number of, let’s say robberies, we’re expecting maybe one or two a day, but we have 10, and I just got an alert saying, hey, there’s a statistical anomaly here. You might want to check this out. That helps public safety because it helps them say, oh, okay, this might be of interest. And the ultimate goal there is to get ahead of it before it spins out of control and also to catch the bad guys in the act. If there’s a way you can connect the dots between incidents and radio out to the field and say, hey, get down. We believe this minor incident is actually tied to something bigger and broader.

JWh: Yeah.

JWi: That’s great. I mean, that what we’re here to do. And I think the key, too, is it cannot be intrusive. It’s just got to be a little alert. What do you want to do with it? Okay. Tells you what the recommended action is. You decide what to do. You can share it. You can attach it to the event. You can even go back and review it later. It’s very discreet because there’s enough crap going on already in the call centre. And the goal, like I said, when this pops up, it should grab your attention. If you want to dismiss it, dismiss it. If you want to open it up to look at it, you can look at it.

JWh: Don’t give me another tool that’s going to get in my way of doing my job is what they’re saying, and this isn’t going to do that. This is going to work in the background, provide you the information, and you decide if you want to do something with it. And I think that that’s a great point to make. So, Jack, as we kind of wrap things up here, let’s talk a little bit to the administrators or to the chiefs. You know, they’re hearing these words, and they’re hearing this now in public safety and how can it be beneficial to them. But, you know, we really haven’t talked about some of the options or some of the ways that you can utilise it here. I mean, these scenarios that we’ve brought up, it sounds like, hey, you plug this thing in and it’s going to magically do everything for you. But they have choices as to what it will actually do, correct?

JWi: Oh, yeah. There’s a configurability aspect to it. So, the agents themselves use publicly-known theories. I mean, to be honest, it’s not like we are mathematical geniuses and created some novel algorithm. But what we did, and most A.I. applications do, is it’s the combination of the different approaches that you package up into what we call an agent that’s specifically built for public safety workflows. But, you know, our whole thing is assistive and being transparent and explainable and what I like to call people-centric A.I. The agency ultimately has control even over the A.I. They can go into the A.I. and say, hey, I want to ignore these particular types of things. We’ve set up ad hoc missions for the agents to go look at. Maybe there’s something going on that week. We want to monitor a specific thing. Maybe you want to monitor a location, or the similarities are always going to be turning out. But ultimately, you can turn them on and off. You can decide who gets the notifications, who doesn’t. The notifications are very explainable and interpretable, which is very important. It’s not like, why did I get this? This is random. So, I mean, that people-centric approach. And then the chiefs and the guys at top, I mean, you see a lot of cities and agencies that are talking about A.I., but they’ll bring up the ethical concerns as well. And we’re very aware of that. And that’s something that we take very seriously.

JWh: Agreed. But I think the important aspect there, and you touched on it, is you don’t have to jump into the deep end of the pool when it comes to A.I. This product will allow you to, I’m going to say, baby step your way into it. And as you see the value, maybe turn on different actions and turn on different filters to provide specific people the information that they require at the time that they require it. So, the chief is going to be able to make that decision. It’s not going to be an all or nothing. It really is configurable to allow them to make those decisions, then. I like hearing that because I think that that helps with the adoption and bringing that in and people accepting this running in the background. You also made the point, too. It’s not all of a sudden going to be all of this information dumping out to everybody. Why am I seeing this? It’s going to be very specific information and detail the way that the administrators want it.

JWi: Yeah. I mean, now, granted, they don’t have to be an expert in A.I. to set this up either. So, I don’t want to intimidate folks. But yeah, I mean, it’s there. It’s an addition to what you already have. All it can do is provide you with nuggets of information that you otherwise would have to have done manually by searching or have through previous experience. It’s going to make you aware of it. And like I said, John, I personally, when we roll this out, I anticipate that they’ll roll it out to a few people first, maybe a couple of dispatchers or one or two call-takers, the supervisor staff, and maybe even someone like an analyst. I don’t think they’re going to just flip the switch, and everybody gets it. But I will say this: within 10 years, this will be table stakes for any public safety application.

JWh: That’s right.

JWi: It’s going to be embedded. So, this was our chance to say, hey, we recognise as Hexagon, as innovation leaders, that this is the future, and let’s start looking at it and not trying to create some sort of pie in the sky, futuristic A.I. No, let’s create a practical A.I. solution that helps our users. And then by doing this, we get our users and the people who buy our systems familiar with the concept, comfortable with the concept. We can develop it out further. And it’s just so exciting because, like I said, I view it as a way to help us, to help our customers, not to replace and automate. And that’s the key. And I keep coming back to that. But that is the key, John. Assist.

JWh: Yeah. It’s just a great tool, I think, for agencies to look in to and have in their arsenal. And what I would say is you don’t have to take our word for it. There’s a lot of information out there on the web. I would recommend, if you’re interested for more information, to go to hxgnsmartadvisor.com. And I think that there’s a lot of great information out there that people can look and see what exactly is John and Jack talking about here. You’ll be able to see some different videos and be able to see some examples of real life and how that tool in the toolbox will assist your personnel. So, a lot of good information out there. Man, you do an artificial intelligence search. I did an artificial intelligence in public safety search, and man, I’m telling you, Google just blows up. There’s a lot of information out there. And I just think it’s a great topic and a great tool. So, Jack, thank you very much. As always, it’s always great talking with you and hearing some of the excitement. I know you’re excited about the A.I. But to hear additional episodes or learn more, visit us at hxgnspotlight.com. And thanks for tuning in.