Speed Up to Slow Down: AI as a Force Multiplier EMs with Tom Sivak

Speaker A: We wrap our arms around technology or technology will run circles around us. We have to be innovative in emergency management. We have to think differently, that disasters are becoming more complex, they're becoming more political. But also at the same time, our community members, because of AI, are becoming more educated on what the emergency management community is doing.

Speaker B: In a world filled with chaos and a myriad of risks, there is opportunity. You're listening to Riding the Wave Project Management for Emergency Managers, where we discuss how we adapt and rise above those rolling waves of hazards and threats we face and rise to the top. And now your host, the president of Pinnacle Performance Management, Andrew Boyarsky.

Speaker C: Tom Sebak is the Chief emergency manager at EM, UM1, where he's focused on putting AI to work for real work, preparedness, response and operational planning. He previously served as The FEMA Region 5 administrator leading federal emergency management support across the Great Lakes region, including dozens of tribal nations, and was also uh, held senior roles in emergency management, Chicago and in Cook County. And uh, Tom, I want to, first of all, thank you very much for coming on the podcast.

Speaker A: Oh, it's great to be here, Andrew. Thanks for having me today.

Speaker C: It was great seeing you at the IAEM, uh, Region 6 symposium and it was great to be able to connect. I know I saw you also briefly. I think we uh, met at the national conference in Kentucky. So I'm finally glad to be able to talk about uh, the work that you're doing at EM1 and its role in artificial intelligence and LLM, which is in the ether it is surrounding us on a regular basis. I want to point out one starting point, the uh, Argonne National Lab report on emergency state of emergency management and emergency management agencies states that they are, as we sort of commonly known, um, we're talking about not the big cities like New York City is not a great example of an emergency management agency that, you know, is, you know, is representative necessarily or some other jurisdictions. But if we're talking about the large swath of the country, counties, local localities, even some states and cities, they're underfunded, understaffed, still expected to deliver. As I like to say, I don't have never heard an emergency manager say, I, um, have all the resources that I need to meet my mission. I never heard that AI is a potential force multiplier. As we talked a little bit a while ago about why is the argument for artificial intelligence one of survival, not just innovation for let's say the one person EM shop or the multi hatted emergency manager.

Speaker A: Yeah, you know, Andrew, good question. Right. And really I love how we're starting this off. So let's, let's set the bar. Underfunded, understaffed, under resourced. We all struggle with this, especially in today's era. Uh, so as we look at emergency management, we have always been wrapping our arms around the latest and greatest technology. With that though, the uh, creation of AI a couple years ago when ChatGPT, you know, really hit, hit the market, there uh, was another tool in our toolbox. Uh, and that's exciting, right? When we have another tool in our toolbox, there's going to be practitioners who are going to be able to like basically embrace it, make it a part of their daily life and then keep moving forward. But one of the challenges we have in emergency management, uh, is that there's, there's always another crisis, there's always another issue, there's always another compliance role and responsibility that we have. And so as we look at where AI is today and where it's going, where I see it is that it is going to be that force multiplier that we're looking for. Uh, I heard the other day where someone said, hey, we have a lot of compliance roles and responsibilities, which means we never get to the list of what we'd like to do. Uh, and there's a lot of professionals that have that list of what they love to accomplish, but they're basically set in the ways of the daily grind of the administrative functions of what emergency management brings to the table. In addition to when that, the, when the incident happens or the uh, weather event takes place where they need to stop what their day job is and move into their grace guy day preparedness and planning. It basically disrupts the apple cart in this life cycle for a long amount of time. So when we look at this in totality, the one thing that comes to mind as you bring this up is that when we look at the world of emergency management today, I look at it as in three levels. First, strategic, right? How many emergency managers today have this opportunity to really think strategically, not one to three years out, but five to 10 years out. As we know that disasters are becoming more complex and, and we understand that the data starts to exist, that's going to help us drive those decisions that we make today in preparation for the next, for tomorrow, but more so in the next five to 10 years. So we have to look at the strategic approach of where we're not always able to get there then the other level that I see in emergency managed and I can assist us with that, they can synthesize that data, it can bring together tightly what our complexities are within our uh, within our agencies based on historical disasters that have taken place and then give us a pathway of how we could start to build that resilience, whether it's in prevention, preparedness or even mitigation, uh, overall. The other side of it too though, right? If we look at it another way, there's this operational side of emergency management. We're in this daily grind of uh, planning for special events, planning for severe weather season, which for many of us now is becoming a 365 day a year challenge that we face. In addition to technological and human induced threats that we have. And in that operational base, what we're trying to accomplish is make sure that we're bringing the right people to the table at the right time to ensure they know what to do when and where. So I remember countless emergency operation uh, center activations that I had like when I was in the city of Chicago and when I was in the city of Indianapolis where I had a new person every single time. And I was retraining that person and they never were able like it was never this thing of I own this role and responsibility, I know my job. And while we retrain them, what they started to see when they came into the Emergency Operations center was how we have to be this coordinator to bring everybody together, to ensure everybody's on the same page overall. So I see that as that operational world and where AI can basically intersect is making sure, synthesizing the situation reports of what's taking place in the Emergency Operations center, being able to then take that information and put practical application to what we're going to do in the next 12, 24, 48 hours overall. So I think that we a lot of time live in this world of operational emergency management. And then that last level of emergency action is more that tactical level boots on the ground, getting out there, uh, when these incidents take place in that coordination role. Right. I, I still feel that we are a coordinator. We are the ones that know that can bring in the right people to be the problem solvers of whatever situation that takes place. But we have to be doing that in unison. And so on that tactical level AI still has that ability to be able to synthesize the information and the data that we have at our fingertips to be able to help, give us courses of action, do nothing, do something drastic. Right? Those are three courses of action that if we have that information in a rapid amount of time, we have that ability to then make the decision of what's best for the community. So I go back to one thing Kyle King told me, you know, asked me in, uh, one of a previous podcast, which was speed up to slow down. And I've taken that and I've really embraced this concept of if we can speed up, uh, the information that's coming in, uh, to us as emergency managers, synthesize that information to provide courses of action, it's going to allow us that mind time to think strategically, operationally and tactically within a certain amount of time when we're focusing on the key components of life safety, incident stabilization and property conservation.

Speaker C: I like that take of speed up to slow down because I have a, uh, close colleague of mine who is a lieutenant in the fire department, and his take is sort of the contrast of that. You got to take time to make time.

Speaker A: Yep.

Speaker C: Uh, you know, and he tells me, you know, even when we're getting on that rig, you know, we get the call, we're, I'm getting my gear on, going down, going down the pole. Nobody goes in the pole anymore. I guess they're not doing that. But you, you get, I mean, you're getting on the rig and you're getting a feed of what's, what the situation is and assessing your sort of line of attack, so to speak. And I think that's a really great point, right, to sort of, we need to accelerate so we can slow things down a little bit and use the tools that help enable us to really assess the situation. Well, one thing I just want to change the channel on here or change our look at this is that for some folks looking at AI, there's, uh, a hesitancy and fear that it's coming after our jobs. We hear this, the news, on a regular basis. Uh, in contrast to that, there's an opportunity. You know, humans still control the levers, so to speak. Uh, the human in the loop, so to speak. Or I've heard recently, slightly different take a human on the loop, so to speak. How would you address that core fear and how emergency managers can wrap their arms around it better?

Speaker A: Yeah, Andrew, I go back, a lot of people have heard me say it. I, I, I stick to my, stick to the way I look at it. When I was at the city of Chicago, uh, when I was leading our team at the city of Chicago, the one thing I sat there and said one day was, we wrap our arms around technology or technology will run circles around us. We have to be innovative in emergency management. We have to think differently. The disasters are becoming More complex. They're becoming, they're becoming more political. They are also, at the same time our community members, because of AI are becoming more educated on what the emergency management community is doing. Right. Uh, an emergency operations plan for a community is usually public information. And if that's the case when these disasters happen, they're going to be more educated on what the roles and responsibilities are as well. So when we take all these components into that situation, my, my concept of why are we hesitant? Uh, is because it's new, it's still new. It was this tool that was given, that was put into our toolbox that we had an idea of what we wanted to do. We know that we needed help with trainings and exercises and public inform, you know, public, uh, you know, the press releases and uh, writing plans, you know, the whole preparedness cycle. But at the same time we didn't really see where those fit into play. And so what, uh, what my feeling in the hesitancy world is that this will pass, uh, where we will wrap our arms around technology is being able to put it into a process. Just like we create plan emergency operations plans for local communities, we have to create this AI engagement plan of how our organizations and how agencies and how we are going to intersect AI into our daily workflow. You know, we uh, you know, you got excited when you heard me say okay, where are we today? Uh, and where we are today is we're in a project based emergency management cycle on AI, where basically what that means is that we're checking boxes, we're doing the compliance, we're getting the trainings done. That's a project to me, creating uh, a plan to do the training, to do the exercise. That's a project. You have a starting point and you have an end point, right? And so we have to manage those projects appropriately. And so that's one aspect that we see a lot of people engaging with today in testing the waters. Where my vision is is that we move from this project based emergency management to a plan program based emergency management of where AI is part of our regular workflows, where it's not used every single day yet, but it's something that we know we're going to use every single time. Prime example, grants. Grants are uh, sometimes where we're writing our threat hazard, incident risk analysis and our state preparedness reports, we're identifying our gaps, right? It's basically capabilities, uh, based emergency management and then knowing where we're going to get those resources if something were to happen on our catastrophic bad day and so in that process we have the ability to ensure that we understand that entire grants process. We have to have the thyroid to know where we're going to uh, put and make our investments, uh, within our jurisdictions. We then know from there how we need to go through the, you know, the uh, poetic model, right, the planning, organizing, training and exercise model to make sure that we are holistically looking at this in a way that's going to be beneficial to that organization. And so in that grant cycle we're utilizing AI in that component. And then lastly writing the grants.

Speaker C: Right.

Speaker A: Basically it's being able to put in for the grants and we see that grants might be, you know, grants are challenged right now. We've seen the reduction in grants, we've seen other areas that have received more money in grants and weren't necessarily expecting it. And we don't know what's going to happen in the grant cycle. It's just kind of a void right now. But that's going to then be another challenge emergency managers are going to face when we aren't keeping equipment up to the standards. Uh, right, because grant funding has been reduced, which means budget cycles. Some budget cycles are a year, some budget cycles are two years. Uh, so we have to be thinking through that process overall. And so as we look through that regular use of AI, that's kind of that idea of where we know that we can utilize it on a regular basis. And then understanding notice of funding opportunities is a big one. Everybody wants to know what's the difference between last year and this year. AI has that great ability to tell us within seconds as opposed to having to reading the two of them. And then lastly, uh, is this daily use of where we know that we are going to use AI in every single thing we do, whether it's a plan and a project, whether it's a training and an exercise, whether it's assisting us in synthesizing the data that's coming in on a regular basis. Daily situation reports, it's this component that we're utilizing it every single day and that daily use is making sure that you're integrating it into your workflows. And now from an emergency manager, I never really thought about workflows, but as I've gotten into this technology world, the workflows are ah, what help us identify where the pain points are to make sure that we can continue to enhance and uh, build our capacity capability along the way as well.

Speaker C: One of the things that I heard you say recently, I don't know whether it was a podcast I don't know if it was a speaking engagement, that plan should come to us to turn static PDFs into living decision support. So the dream isn't AI necessarily writing the EOP, although in certain instances it has some capabilities to be able to do that or help revise it, given new information, data, et cetera. As you just mentioned, the rapid synthesis of all of that, it's really AI turning the plan into usable role based guidance at, let's say 0200 during an activation. Can you provide and explain an example of this and how it applies for our audience?

Speaker A: Yeah, absolutely. I mean this is my vision, right? This is like if someone were to say, Tom, what would you love? I would love for us to be proactive and not reactive. You know, we are, we by nature are a reactive group and we do take proactive measures, but usually that's when we're notified that something's taking place. How many of us, when we have been notified that we're going to have a severe weather outbreak within our jurisdiction, go to our plan, pull the plan off the shelf, review what our plan roles, policies and procedures are, bring everybody together, have the consequence management call, and then ultimately make sure that we are staying in contact on a regular basis. When I go back to my time in the city of Chicago, uh, we had to be proactive. Failure was never an option. It was ingrained to me from my leadership from day one that we have a role and responsibility within the city. They made sure that every single time, whether it was a disruption that could have had, uh, implicating traffic components on, um, you know, disable Lakeshore Drive to, you know, large fires that would have uh, an impact on citywide operations. It was ingrained in us to always be as proactive as possible overall. And so my vision that I have is that when we get notified that we have and weather is an easy one, when we get notified that we have a potential severe weather outbreak, the storm Prediction center is providing that information sometimes five to seven days ahead of time. And if it's telling us, hey, you're going to have a severe weather outbreak, we don't know how bad it is, but it's going to have a high inch rain event, it's going to have high winds, it might have other implicating factors such as power outages. The idea that I have is that benchmarks off of our plans, so our emergency operations plans and it actually operationalizes them before that disaster even takes place. Now it might turn into be just an exercise, you know, the weather the weather Service sometimes has those forecasts and they might be wrong. And we've all. And, uh, you know, I think some of the most professional people to say that they were wrong is actually in the National Weather Service and they say, hey, you know, sometimes the forecast models it was something different. But if it basically says, hey, you're going to have a severe weather outbreak and you don't, you at least prepared appropriately to make sure everybody was on the same page. They knew what their roles and responsibilities are and they were ready to respond should an incident happen. But if that incident happens, what people will start saying is, hey, we talked about this. We had the consequence management call that, uh, ensured that we were on the same page. We knew when we were going to be getting regular situation updates of what the situation was and what the implications were across that jurisdiction. And we knew that if something were to happen, what our activation point was to then go into the Emergency Operations Center. So instead of the words, uh, we never thought about this, we didn't think it would happen to us, we were ready, but we didn't necessarily know where it was going to hit. That terminology changes with the advent of AI of um, we planned for this, we trained for this, we exercised this, we discussed this, and we know exactly what to do now in the next 12 hours, 24 hours, the next 30 days, depending on the, the incident in and of itself. And everybody just keeps moving through that process so they can get back on the road to recovery as quickly as possible. Because that's the ultimate goal, making sure that communities are more resilient, making sure the emergency management industry is more resilient and they deal with more just stressors as opposed to the shocks and to be able to then focus on the communities that are served. Because in the end of the day, Andrew, it always comes back to people. That's what really matters most.

Speaker C: And I think there are a couple things that I just wanted to bring out, and that is most of our major disasters are either hydrometeorological events, very fancy word for storms, and a lot of rain and wind in one place. I always like to say in my classes that I teach academically, and then major fire events, heat, extreme heat, those sorts of things. There's a lot in between, but those are some of the major sort of. And you know, of course floods are the other consequence that takes place as well. Uh, given that and the National Weather Service's ability now for things like that, I saw presented at the national conference worn on forecast, for example, for tornadic activity It's a game changer to be able to predict hours in advance war when potential uh, you know, stor tornadic uh activity is going to be present and where it will impact uh, for emergency managers. So that, that kind of thing is that, that's the kind of thing that we're talking about in terms of being able to look at and to create a greater degree of certainty in terms of decision making.

Speaker A: Oh, 100%. And Andrew, I like, I see it like this, you know, right when we hear, when we get these forecasts, what do we do with that information? We can do nothing, wait and see. We can do something of looking at mitigation measures before that incident takes place. Right. When I was uh, at the city of Chicago, as soon as we knew we were going to have a high inch rain event, there was a particular neighborhood that we knew always flooded. And so what we would do is we would get notified by the National Weather Service and we had our game plan. We actually had the Albany park flooding plan and that was get in touch with streets and sanitation to be able to street sweep all the streets. And then another one was clean out all the basins. So get with the water department to clean out all the basins. The other one was work with the all uh, the person's office to be able to move cars that might be near the river that could potentially flood. Right. So we had this activation plan and we knew every time that we would get a certain flooding event and hit that trigger. We had to do that every single time. And then it got to the point where there are times when we actually put staff out in the field. But that was, we knew it was going to take, we had an inkling it was going to take place. We were instructed it was going to take place. We had the plan for it and therefore that was the execution point on it. And so the more times we can do that where the plan says you have a high inch rent event that's going to happen in the next five to seven days. Here are some mitigation actions to consider. Uh, right now, uh, you should consider in the next two days a consequence management call to make sure that all agencies are on the same page. And then 48 hours or 24 hours ahead of time, ensure there's regular situation reports that are provided uh, to agency personnel. Everybody knows what the game plan is. No one can say they were taken off guard. And then in the background what are we doing is we're making sure that we're making the phone calls, right? We all live on this, right? We all live on the phone, but we're all making sure that the agency heads are making sure that they are activating their procedures appropriately to make sure that we are taking care of as much of the community as possible with what we have at our fingertips, which is that information. And really is how do we take that information and do we turn it into more intelligence, actionable intelligence? Do we use it as a precipitator to make sure that we are all working in alignment overall to focus on the community? That's what it comes back to.

Speaker C: So one thing I want to come back to is what we started you mentioned at the beginning of the podcast, which is the velocity of the crisis that we could be facing with different types of disaster events, cascading, compounding event, uh, impacts, and also the information overload from all the data, social media communications, etc. That's coming in. So AI could be a triage for our human brains, which have a limited capacity and being able to take and synthesize this. How can emergency managers though, protect decision making from the potential bad inputs, misinformation, disinformation, incorrect data, so to speak, as well as incomplete outputs when AI can sometimes amplify errors, as we know, sometimes AIs can hallucinate, take data, put it in there for some reason when it doesn't fit, that sort of thing. So how do they protect against the inputs and outputs aspect?

Speaker A: That's a good question. Uh, so a couple, uh, there's a couple routes I want to go with this. The first thing is we as practitioners have years of experience that are within this realm, right? We have, we have mentors and coaches that have been a part of our careers, ah, as we've traversed our careers, right. So one thing I think about is that we are in control of anything that's outputted from AI right now. I always say that because we don't know what the future is going to hold. And there's a good chance the future might have more accuracy than we do as humans. But for right now, we are still in control of it. We are the ones that are putting the prompts into the systems to look for those outcomes. We are the ones that should know what our plans say and what our policies and procedures say. And we are the ones that should be taking that information and using it as a guide, uh, as a consideration of action. Uh, and so the way I look at it is trusted information. Uh, so right now these emergency operations plans that have been created, even if they're older, have been validated, they've been worked through, they've been trained, they've been exercised on. And so we have a good amount of plans that have been created that people know of. Okay? So because of that, there's a starting point that has information that we know exists. Additionally, as we look at trusted information, we have to know who the trusted sources are. National Weather Service is an easy one for us because it's the only alerting authority in the nation as it relates to natural hazards. And as we move into more technological and human induced hazards, of course, that information changes and we have that reliance, especially on human induced fusion centers. And you know, really, if we go back to it, the latest Comprehensive Emergency Planning Guide 101 from FEMA talks about this integration of fusion centers and engagement in emergency management. And I believe that that's a good thing, uh, because of course we have to be thinking about all hazards, whether it's human induced, technological, natural hazards overall. But I think about with weather, it's an easy one because we only have one alerting authority. So trusting that alerting authority, and we do know that there's many jurisdictions now more than ever that uh, might have contracted meteorologists, they might have their own meteorologists on the inside. And being able to take all that information, synthesizing it to make sure that they're focusing on what's going to be best for the community is important. AI does not have feelings yet, right? And so we have got intuition, uh, we have to be thinking about, okay, well, what's the worst case scenario? But if we're able to synthesize that information so quickly, it gives us more time to think about the what ifs, to ask our what are we missing? What does opportunity look like? We're always things that I brought up, whether I was at the city of Chicago or at FEMA Region 5 was making sure we left no stone unturned. I always ask three questions to our team. Who are we missing? What are we missing? And what does opportunity look like?

Speaker D: Like?

Speaker A: And that was always the conversation that we'd have as we were closing up a meeting. Because what I wanted us to do is keep thinking, what are we missing? Because ultimately every single disaster I've gone into, we've missed something. And usually I go back to my time in, uh, the city of Chicago. There was always an agency that wasn't at the table, that wanted to be at the table. And we had to evaluate are they going to be a value add? Should they just receive that information? So then they're ready on their end to be able to respond to anything within their jurisdictional boundaries. And uh, and so I always have to keep that in mind, uh, as we work through it. The last thing is with the hallucinations, you know, really it's a matter of how you utilize A.I. uh, and being willing to receive the hallucinations and knowing the answers to the test. Uh, I remember early on I put into one of the applications, give me, uh, an overview of every single disaster response that's taking place in FEMA Region 5 in the last five years. And I knew every single one of them and the output was complete and utterly wrong. And I knew it though, because I knew every single disaster response that took place. And so knowing the answers to the test or knowing where you get to get the answers to the test is going to be what's going to be helpful overall. And really, when I think about it, right, it goes back to the hesitancy conversation as well. This is as bad as it's ever going to be. And so a year ago it was as bad as it's ever going to be. It's better now than it was and it's only going to continue to grow, to be a part of our lives, uh, to a degree. But at the same time, many people will say to me, Tom, I love this. I love, uh, I love what you all are doing. I love that you're engaged in this. I'm scared I'm going to lose the ability to be a planner. I'm scared that I'm, that uh, people that I work with are going to lose this ability to go back pencil and paper. And this is where our leadership comes into play. It is our responsibility to make sure we know how to plan, that we teach people how to plan the proper way, that we teach innovation, but also at the same time that we go back to what we are in this worst case scenario, which means go back to pencil and paper, don't stop the exercises where you pull the Internet plug and you have to be innovative, uh, on how you're going to engage, uh, with your agency partners and then also your statewide partners and your federal partners, that's on us. And so we need to keep that at the forefront of it as well. But if we have more time to synthesize this and be able to create our plans, policies, procedures, trainings and exercises, it will also allow us to make sure that we have a, uh, track that's focusing on going back to the basics and making sure we keep the basics at the forefront and our foundation of what we do every single day.

Speaker C: Well, I want to thank you again for coming on the podcast. This has been a great conversation. Tom Sebek is the Chief emergency manager at EM1 uh, where he focuses on AI for real world preparedness, response and operational planning. Previously served as The FEMA Region 5 administrator leading federal Emergency Management Support across the Great Lakes region, including Travel nations, and he's also held senior roles in Chicago and Cook County. Uh, thank you once again, Thomas. Great conversation.

Speaker A: Thanks Andrew. Looking forward to more In a world

Speaker D: where emergencies are becoming increasingly complex and varied, the need for effective coordination and planning has never been more paramount. Enter the Project Management for Emergency Managers workshop. This unique online learning experience is designed to equip you with the tools and strategies to streamline your emergency management efforts. Over the course of six engaging weekend sessions, we delve deep into the principles of project management tailored specifically for your needs as an emergency manager. Learn the skills needed to navigate the complexities of M emergency management.

Speaker B: You've been listening to Riding the Wave, hosted by Andrew Boyarski, President of uh, Pinnacle Performance Management and Clinical Associate professor in Emergency and Project Management at NYU and John Jay College.