The Alerting Authority

AI Taking Over? How AI is Changing the World of Alerting

• Jeannette Sutton and Eddie Bertola • Season 1 • Episode 37

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 59:58

Artificial Intelligence is rapidly changing the way emergency managers, public information officers, and alerting authorities communicate with the public. But should we trust AI with emergency alerts and crisis communication?

In this special panel discussion, Jeannette Sutton, Eddie Bertola, and Dan Stoneking explore the growing role of AI in emergency management, public warning systems, and crisis communications. Drawing from research, field experience, and communications expertise, they discuss the opportunities, risks, and realities of integrating AI into emergency alerting.

Topics include:

- Generative AI vs. traditional AI tools
- How emergency managers are already using AI
- AI-generated emergency alerts and warning messages
- Trust, credibility, and human oversight
- The future of public information officers
- Message design, warning science, and alert effectiveness
- Why AI should be viewed as a tool—not a replacement

Whether you're an emergency manager, PIO, researcher, dispatcher, or public safety professional, this conversation offers valuable insight into one of the biggest technological shifts facing the alerting community.

#EmergencyManagement #AI #CrisisCommunication #PublicSafety #EmergencyAlerts #IPAWS #AlertingAuthority #ArtificialIntelligence #PIO #EmergencyCommunications

SPEAKER_01

Hi, I'm Jeanette Sutton.

SPEAKER_02

And I'm Eddie Bertzola.

SPEAKER_01

And we welcome you to another episode of the Alerting Authority. And as always, we invite you and encourage you to subscribe, follow, listen, and most importantly, participate in these podcasts. We want your questions, concerns, ideas, problems, and pain points, and your success stories. So we as alerting authorities can do our jobs better and make our communities safer.

SPEAKER_02

That's right. And just want to thank all our sponsors for helping us out and for all of you too. So feel free to uh click on these things, to like them, to share them, to subscribe so that you can hear all the latest updates. And we're actually sponsoring ourselves on this one. So this one is sponsored by the Alerting Authority. So emergency alerts are way too important to leave to chance, outdated policies, or training that actually never gets beyond the PowerPoint. And the Alerting Authority helps organizations strengthen every part of their emergency communication program through governance reviews, policy and procedure development, the all-important message design, hands-on training exercises, and practical guidance built around real-world operations. So whether your organization needs a complete program overhaul or simply wants to make sure that the right people are prepared to send the right message at the right time, well, the alerting authority helps turn plans on paper into confident action when it matters most. So learn more at the alertingauthority.com.

SPEAKER_01

So today we're doing things just a little bit different, differently. Normally Eddie and I interview a special guest who is a subject matter expert, and we pepper that person with questions and it's a discussion. But today we're going to do this with a new subject or subject matter expert who provides a different perspective that we want to actually have a discussion and more of a panel discussion. So that's what we're going to do today. And we're going to focus on the topic of AI. And we each bring our own expertise, our own disciples from our own disciplinary backgrounds. And I think this is going to be a really fun conversation because we each have very different perspectives on this. So first, um, I want to introduce to our audience Dan Stonek, who is the owner and principal of Stoneking Strategic Communications, the author of Crisis Communications and Emergency Management, the founder and vice president of the Emergency Management External Affairs Association, the founder of Message Prism, and an adjunct professor in the communications department at Westchester University. Among his myriad deployments and lifelong experience as a public information officer, he has supported the response to the Pentagon following the attacks of 9-11, was the lead spokesperson for the National Guard in Louisiana during Hurricane Katrina, responded to the 2010 Haiti earthquake, and deployed to Puerto Rico in the aftermath of Hurricane Maria. He also served as the External Affairs Director for FEMA Region 3, leading his team through the 2015 Papel visit, the 2016 Democratic National Convention, and the response to the January 6, 2021 attacks on our nation's capital. So we welcome Dan to participating in our show today and being part of our panel. And I also want to introduce my co-host, Eddie Bertola. Eddie is the founder of the Alerting Authority and brings the practitioner's perspective to emergency alerts and warnings, the place where policy, research, technology, and public expectations finally have to work in the real world. With nearly two decades in state and local law enforcement, experience supervising a statewide emergency operations center, managing major public alerting programs, and training emergency professionals across the country, he focuses on what happens when the discussion ends, and someone actually has to make the decision, write the message, choose the tool, target the right area, and push the button. He also holds a master's degree in organizational leadership and is produ pursuing a doctor of strategic leadership, giving him a balance of education and boots on the ground experience. Or more simply, he understands the theory, but he also had to make it work when the clock is ticking. Your third panelist is me. I am a social. I am a social scientist. I'm an academic. I have studied alerts, warnings, and emergency risk communication across all channels for nearly three decades. I'm employed as an associate professor at the University in Albany, and I'm the founder of the Warn Room. My research has been funded by the National Science Foundation, NOAA, USGS, and FEMA with my most recent project funded by the FEMA IPAUS program, the integrated public alert and warning system, where we built the message design dashboard and conducted studies on effective messaging and developed training materials. And I also serve as an alert and warning subject matter advisor to EM1, a purpose-built AI tool for emergency managers. So, listeners, you could tell we have three very different perspectives that we're bringing to this topic. And I am really excited to welcome Dan in particular to the show. And we're going to kick it off with him asking from each of our panelists to share a little bit more about your background, your the area of practice that you have served in and the different experiences that you've had. So can you describe a little bit more about your professional background and where alerts and warnings fit into your research or practice? Dan?

SPEAKER_00

Yeah, of course. So Jeanette and Eddie, first, thank you very much for including me. This is an important and exciting topic. And all of the things you listed about titles and experiences just mean I have gray hair. So that's that's what I got from that. But I like what you said, that we all have different experiences, right? So mine uh primarily come from the perspective as a crisis communicator, but I've had that view both at the federal level of FEMA, at the regional level uh uh with FEMA also, but also in relationship with our states and locals because we interact so often. Uh but I've also had perspective as a journalist. I I've written for an Oxford paper with some colleagues, and I write for Homeland Security Today, where I address this topic. And that that gives you a second way to look at things because I'm asking more questions, the hard questions, um, as opposed to working in just the operators. So thanks.

SPEAKER_01

Thanks. Eddie, where where do alerts and warnings fit into your practice and research?

SPEAKER_02

I would say I am just like it said on there, uh, where the rubber meets the road, where I mean, when alerts actually have to go out, me and my colleagues are the ones that send them. We're the ones that um have to translate whatever the policy is telling us, whatever the hazard is requiring us to um talk about and translate all that into an actual alert where we have to know how to send it, where to send it, when to send it, um, and how to deal with all the potential fallout with everything that can come with. Um kind of like Dan, I have a lot of respect for PIOs out there. Uh, me and my colleagues, I would say there's there's a sense of courage um in law enforcement or being a first responder. You run towards danger and yet you run away from cameras and media. Um and so I I'm really excited to get his perspective on there too. And Jeanette, too, I didn't know if you wanted to introduce if we have our fourth panelist, if Jingles is around. I tell you, people, if you were watching this on YouTube, you normally see he's got to be the official sponsor or official uh official animal of the alerting authority because it's fun when he makes an appearance. So in case you see him, I'm sure he will, you know, have to come on camera just for a second.

SPEAKER_01

I'm not sure where he is right now, but we'll call him our podcat.

SPEAKER_02

Oh wait. I've never heard that. That's awesome.

SPEAKER_01

Well, I'll let me share a little bit about where alerts and warnings fits into my background. Um, as a researcher, I've spent most of my career studying this topic, um, looking across different channels, the ways that messages are shared, disseminated, amplified through social networks, the kinds of contents that make a difference to make those messages go farther. And then with my research team, we've done surveys, focus groups, interviews, experiments, eye tracking studies, um, all the different strategies that we can use to try to understand how individual members of the public interpret the warning message that they receive on their cell phone or as a some sort of a digital message as a wireless emergency alert or as a social media message. And how do they interpret that? And does it motivate them to take action? And our objective is always to determine how we can communicate more effectively so that people are they they have less delay in their actions and are motivated to act more quickly. And then in in addition to the research, I then translate that research for practitioners to the best of my ability so that practitioners can help to implement our findings, uh, which is what we did with our IPAUS funded uh project, where we developed the message design dashboard and published the warning lexicon, which is available freely to anyone to download. It's to just look for the warning lexicon. 18,000 downloads so far, domestically and internationally. So please grab that resource.

SPEAKER_02

I think we'll even put it down below, right? So if you're here, if you're watching, you should be able to see it.

SPEAKER_01

Absolutely. Um, all right. So let's jump into our conversation about AI. Um, you know, I it feels like we're in this real technological shift, and AI is emerging as either like an adopt it or be left behind tool. And I remember when I started studying social media with NSF funding around 2005 to 2007 in the early 2000s when social media was becoming a thing. And we had Wikipedia, MySpace, and Flickr early on, and then we saw the emergence of Twitter and Facebook and YouTube and then Instagram. And there was so much resistance to these platforms. And then we saw this change after 2010, and they became adopted by federal agencies. The National Weather Service adopted Twitter for the first time in 2012, the National Hurricane Center in 2014. Um, I remember teams working with FEMA in response to the Haiti earthquake in helping to pull data together online, and it was suddenly recognized as this useful tool. Um, and I feel like that's where we are with AI right now. It seems like it's everywhere, but there are significant significantly mixed feelings about adopting it. So, does this historical comparison make sense to you? And what are your thoughts on this?

SPEAKER_00

So sure, I'll jump in. Uh I I love the you you you ended with the mixed feelings, right? Because that's nothing new. Like if we look at the history of time, um, people were afraid the Gutenberg uh press would eliminate anyone ever going to see theater anymore and books for the thing, then radio, then TV, then the internet, and then AI. But when we pause, they all still exist, right? So they didn't replace each other. The challenge is artificial intelligence is moving things exponentially faster. Um, we only found the internet a minute ago, and a second ago we found AI, and it's continued to move fast. Um but because of that, I think it's important to jump on and to learn and accept that there's been a change in the history of how we communicate and how we process information. Uh, and I like to give the example of where VHS tapes got replaced by DVDs, got replaced by streaming. And in that example, uh Blockbuster didn't make that third and final step and went out of business. But Netflix did and they prospered, right? So that's just one historical example of how choices are made result in good and bad consequences. And and I just want to finish by saying, you know, things in life can be good and bad, right? Um, that's not uncommon. And I think that the best course of action with artificial intelligence is separating the wheat from the chaff. And I think we should work to regulate the bad and leverage the good.

SPEAKER_01

Yeah, thank you. I I love those comparisons between all these much, much more historical events and how we have seen this over and over again. There is resistance to new technology, and then it becomes normalized, and everyone adopts it. But you know, there's there is there are significant questions right now among a lot of people. Eddie, what do you think?

SPEAKER_02

I think people are resistant to change overall. Yeah. And one thing that helps us with change is when it's done slowly, methodically, and we have a chance to put in our thoughts and our opinions. And um, I think that helps. But with AI, just as Dan said, it it's moving like there's dog ears, there's but then we look at computers, you know, and technology seems to double its capacity every few months. And then AI has taken all of those things and just as he said, has accelerated it so quickly. Um, I am a proponent of the the positives of it. And I think one of the things for this, just like social media, and for anybody in so my generation, that's one thing that we learned the hard way in law enforcement as a first responder is jumping onto that bandwagon late. I mean, we we lost so much potential um by no, we're not gonna go social media. Nope, nope, that's wrong. We can't control it, all these other arguments that ended up being, you know, just we we just weren't ready for it apparently. Um we were ready to take the leap. And my fear is that we are in that same boat when it comes to AI technology. And I know we're gonna get into what that all means here in just a little bit. So my my hope is to get involved. I I can't even say early because it's not early anymore. Things have been moving quickly, but get involved now so that we can put our thoughts and our opinions and where we hope it goes or it should go out there so that you know we can be more directing it versus just trying to play catch up.

SPEAKER_01

Yeah, yeah. Well, speaking of definitions of AI, yeah, we're using that term, but it means a lot of different things to a lot of different people. So when you when you say AI, what do you mean?

SPEAKER_02

Um if I can hit one part of AI that I think is what we talk about the most in terms of being scary, um, is generative AI, something, you know, it creates new, unique content. Um it's it, you know, for any students out there, it's potentially writing your papers, which it shouldn't, right? But that that idea where AI is um generating information that we did not give it in advance. So it's no longer just processing mass amounts of information, helping us to see um gaps and an analysis we've done, or help us to combine so much information into something that's easily digestible, it's actually creating thoughts, ideas. And I think that's kind of the scary part because you know, we've seen it in the movies, what happens when computers take control, right? Um, and for whatever reason, that's what people are are associating. So I think AI is a term, folks, it's been around a very long time and it's helped us out immensely. Um, whether it's through, you know, your shopping, through medical technologies, I mean, a lot. But the generative thing, at least in my opinion, I think is what is scaring people.

SPEAKER_01

Okay. And Dan, what are you what do you mean when you say deaf um AI?

SPEAKER_00

Well, I I agree with everything Eddie just said. He laid it out very nicely. I I think it might be helpful to add a few more examples of what is not generative AI so people can understand how much it's part of their daily lives. So one example is artificial narrow intelligence. That's one of the many types, and that's what allows my girls to play Roblox on their phones. And and not a lot of people are afraid of that, right? There's limited memory, that's what Google basically is, and we haven't been afraid of Google. And then there are reactive machines like spam filters that can do a lot of good and help for us. So, you know, it's it's just one of many types. And in society, we tend to loosely combine things. Like uh, people all know that Kleenex is tissue that people say, hand me a Kleenex, even if it's another brand, right? So, so having that bigger experience can do a few things to help us understand. One, one, it's an evolution. Two, it has a different aspect. And three, they should all be treated, examined, analyzed, and regulated and leveraged in different ways. So uh generative AI has so many examples, and now whichever person has it first thinks it's also equivalent to their tool. In other words, people say AI and ChatGPT as if they're synonyms, but there's Claude, there's Gemini, you know, there's Copilot, so there's other versions of it. And I think it just helps sometimes to step back and not oversimplify.

SPEAKER_01

Mm-hmm. It does feel a little complicated. Like I'm not sure that I know if a chat bot that pops up when I'm searching a web page and it says, Hey Jeanette, can I help you? Is that generative? Or is that just a large language model that's pulling from the list of most frequently asked questions and answering with uh information that it knows is the correct answer. Um, and maybe if it's generative, it's coming up with a brand new answer. I don't know.

SPEAKER_00

So if I could respond to that, because that's such a great example. So chatbot actually happens to be part of the litter lit, excuse me, the limited memory, like Google. Um, but the nice thing is we don't need to memorize which each one of the different tools comes under. We just need to understand, I think, conceptually, that it is, you know, going back to what Eddie said, it's impacting our lives in so many ways. It's not new. Some of these have gone back to the 1940s. Um, and that I think puts us on a better pathway to examine each tool individually and not make broad assumptions about the overall capability.

SPEAKER_02

I think one of the things that I think worries people the most, maybe not, you know, it's that flashbone or that snapshot of where we are today, it's where we headed with it. And because some people are like, okay, I'm kind of comfortable with this. Um again, there are different generative models um out there. Chat GPT is one. Um, and people have fun with it. But it it's that idea of if this is where we are right now, where are we going to be in five years, where are we gonna be in 10 years? And I think that's what's causing the fear, just because of the unknown. And that's why you see legislation, that's why you see um local governments or even state governments who basically say, we cannot, we're not using AI technology, we're not using it. And it's funny because it's just as Dan said, I want to go over there and whisper in their ear and be like, you've been using AI technology for decades, my friend. But however you want to spin it, that's fine. But but it's that fear and worry.

SPEAKER_01

Yeah, well, I think when I speak with my students and have to help them to understand where the boundaries are in the classroom on using AI, um, some of them are really very savvy and they say, Well, what about um my spell checker? That's AI. What about the systems that help to check for grammar? That's AI. Um, what about a search window in Google? Um, that's AI. So, you know, if you can, if you really get into it, you can see there are so many different applications that have been running for a very, very long time. And those boundaries around what is appropriate use. It seems like it usually does get into these questions around. This agentic AI where something is being created and there's less of the human involved in the creation of that information. And so I think that leads us to the next question, which is, you know, there has been resistance to technology in the past, and there's resistance now. So where are you seeing resistance in what domains for what tasks? And are there places that are suited for generic AI, that focused versus focused AI? Um, you know, and thinking about the different kinds that are out there. Dan, you want to go first?

SPEAKER_00

Happily. Uh so um, you know, I'll speak from my experience, right? Which is crisis communicators, the PIOs. Yeah. And and we're a unique breed when it comes to this because it threatens the very ethos of what we do. Oh my, it's gonna write a press release for me. It's gonna do this task. No, I'm the writer. But uh when they think of it that way, they miss the examples you just gave a moment ago. They use Grammarly, they use spell check. And uh, but there's great resistance. And some other reasons in in our particular group that we resist it is it's not just, well, my ethos and that's my stuff. It's they're concerned about credibility. If somebody knows I used AI, then they think, oh, this is some random computer talking and I'll lose credibility. They also fear job security because, oh, well, if you can get AI to do it. Uh, but the truth is, uh, I think all of those concerns can and have been satisfied if we just look at them appropriately. Uh we will always have job security because you always need the human touch. Um AI can craft it, but they can't fully put the humanity to it. Um and by the way, the bosses don't want to do those final edits. They need the PIO or the communicator to make those uh for them. And uh and credibility, I think, is managed by continuing to do what we've always done, tell the truth. So if you use an AI image to make a point, then credit AI for the image. But if you use AI to uh do a first draft or just get you some citated research information, you don't have to tell the world that you used it to create your final document any more than you said, oh, I should tell you I used a dictionary, a spell check, and grammarly before this brought to you. So I think it's that whole process we're in of the fears we have versus the opportunities we have and how to manage all that successfully is the challenge that our group is really facing.

SPEAKER_01

What about you, Eddie? Where are you seeing resistance?

SPEAKER_02

Well, thankfully, I would say a lot of people in my area, I mean, they've had to in the last 40 years go from a toggle switch in their patrol car to control lights and sirens. I mean, really just this click, click, click, you know, three options to beautiful touch screens that not only control the different light settings, you know, direction, um, duration, um, brightness, um, to all your radio, I mean, everything. And then they start integrating, you know, voice activation. There, there's been a lot of change over the last 40 years. I think there's a lot of um acceptance when it comes to what we're talking about here with the use of AI for for a couple things. One, it's helping us to process information so quickly. Um, and that's one thing that that we need in our field, right? Let's say if we're if we're going over alerts right now, it's we have all these different inputs of information coming. And I want to make sure that it is the most accurate as possible. Not only am I trying to listen to the radio, but there are reports coming in from social media, there are reports coming in and are they verified or or not? And using different tools that exist, we can take all that information and try to vet it and verify it and then use it. And it is so much faster. And it allows us to put more accurate information and more in a more timely way. Um, so I think there's a lot of benefits that way. But it really depends on how you sell it and and how you're talking about it. Because if you say, well, I had, you know, AI craft this or AI gave me this information, because of the issues that exist with the term AI in some areas, people are going to be naturally very resistant. But if you were to say, hey, I use this specific software, well, people are they're okay with that because we use that software. Um, but they don't understand that what's driving that software are all of these AI enhancements.

SPEAKER_01

Yeah. And it Dan, I'm aware that you've helped to create a focused AI specifically for public information officers. Can you tell a little bit about why that is a tool that should be less scary and have less resistance compared to the more generic approach of asking Claude to do something for you?

SPEAKER_00

Yeah. So um it first of all, it gets back to what I said at the very beginning that that we're moving forward at an exponential speed, right? So for businesses, for careers, for professions, I think the the more commonplace AI tools like Claude and ChatGPT are only going to get you so far because they're not experts in your profession. So we built a tool called Message Prism where folks can put in an incident report, an op order, any information about a disaster or crisis. And within seconds, it'll create the first draft of 10 different communications products, like a press release, uh talking points, social media, and more. And why that's different and important is because it's focused. It was designed of, by, and for uh over 100 years of emergency management and crisis communications experience. So ChatGPT doesn't have that. It doesn't have it all combined in a way that that particular career field needs it and is learned through incredible lessons learned and very finite and nuanced needs. And that's just my field, right? It's happening to logisticians who are learning to work with very focused AI tools that know all about logistics, all about uh um water and tarps and all the different types of products that they need to move from point A to point B. And I'll close by saying focused AI tools uh basically will take thousands of pieces of information and hundreds of questions you might ask a chat GPT and put it all in one program to make that work even faster. And it's a critical step forward in the evolution of AI.

SPEAKER_02

Yeah, if I can give an example, Jeanette, it's and I and I talk about this because I I like AI technologies and I'm trying to learn and adapt and apply them. And not every time it's a success. Um, but the tools that I'm currently using, I train them. I put effort into it. And so what I tell people, it's like it's like getting a dog and you go to a dog park, right? Um, or you're out walking your dog and you see this one dog that is just, I mean, it doesn't even need a leash. It stands right next to its dog owner and just walking around, very good with people, can sit. I mean, it's just the perfect ideal dog. And then you see this other one that, I mean, it's the dog taking the owner on a walk. I mean, the they're trying to pull back as much as they can. There's no control. And there's a difference. And people say, well, what's the difference? Well, you put time and effort into training. And so that dog is now responding very well. And AI is no different. And so for me, in the different tools that I use, I've put up parameters, very strong walls to say, this is the information I want you to search in. This is, you know, you're not allowed to go outside here. You're whatever it is based on the project. And it ends up being um pretty effective. And just like Dan said, it's nothing is ever, I've never seen an AI product that's 100%. Um, and I don't know if we'll get there. We may, but it's not there today. But it at least puts forth that initial draft, that initial idea, and then we can build off of that. And then the more that I put into this designated space where this AI assistant lives, where I know the knowledge base, I feel trust, it's a trusted source, whatever it is, it just produces better and better products. So it takes training and it takes effort. If you were to like Jeanette, just for fun, um, try to go on chat, and I think we've done this in some of our training, and just throw in, hey, um, write a message for you know whatever it is. Um, we've seen some pretty crazy and interesting responses. Um, but if you, you know, create a tool that has all of your research and information that people can use, just like the MDD, right? That more messages like that are going to come out because you've put in the time and the effort.

SPEAKER_01

Yeah, yeah. So um when we one of our early trainings that we did in the county, um, we remember that there was uh we asked during an exercise that people are at a table, write a message. And one table finished very, very quickly. And I went over and I read what they did and I said, How how did you do that so well? That was that's great. And they said, Oh, well, we use Chat GPT. And I said, How did you know to trust ChatGPT? And they said, Well, we used all of your materials. And they had the skills and the knowledge and the ability to train that system to say, use the warning lexicon, use the post alert lexicon, and build the message using those, uh, that background, that knowledge. Um, and honestly, when we built the message design dashboard and then and when we were creating all of the data backbone, which is in the warning lexicon and the post-alert lexicon, um, we knew that people who were good and smart and uh willing to adopt AI and program it, train it, that they were going to take those documents and use them to build AI agents that could create an effective alert and warning. Um, and so I, as we were building it, I was resistant because I really wanted selfishly to for people to use the message design dashboard. Um, but we also knew this is this is the evolution of it. It is going in that direction. And to be totally honest, I am having less resistance now than I did before, partly because I've been thinking about AI for months as I work with EM1, but also because when we look at the messages that go up on PBS Warren, they still are consistently lacking the contents we know need to be in a good warning. And if AI agents can create messages that conform to what we know should be in a message, it may be that they become better at creating that first message in comparison to the human who has no training and is sitting at that dashboard. Um, and that first draft could be complete to start with. And then it's a matter of the human getting involved in editing to make sure that all of the information is accurate from the context in which they are sitting in and not just accurate in terms of writing it correctly. Um and so I am I'm positive there's focused AI tools that are coming out. Probably many of our listeners have used some of our materials already to build their own AI agents within their organizations. Um then there are others who are still waiting to figure out our, you know, is this gonna be safe or not? And um, you know, one of our projects with EM1 is to build benchmarks and to help to compare these different uh large language models to see how well they can perform on creating messages for different scenarios. Um, and that's gonna answer some questions about whether or not these are tools that can be utilized effectively for alerts and warnings. So it's a it's a wild world out there. It's changing so fast. Um, and my resistance is is is breaking down.

SPEAKER_02

Well, I want to take a quick moment here. And Jeanette, I need you to give an example, not an example, tell us exactly where to find some of this stuff because and I I know the answer, folks, but I want it because it's hers. She needs to say it. When you're training your models, okay, getting the right information is critical. Again, the first documents I put into mine, right, were the warning lexicon, post alert lexicon, and then the missing person information that they did like Jeanette, for someone who's trying to not avoid, you know, but using the MDD, but if they're trying to do what we're talking about, where can they find those documents?

SPEAKER_01

They can go to the warn room and under the resources tab, they can find the warning lexicon and the post-alert lexicon. We are working on a missing person's lexicon. Um, and that will be posted to the warn room as soon as we've got that finished. Um you can find all of my research papers under my personal academic website, which is Jeanette Sutton.com. And all of that information is also very useful. Um, but the lexicon documents are the most useful because it gives you the the hazard impact statement and the protective actions that go with every one of the messages that we've uh every hazard that we've looked at to build um the the MDD around. So there's more than 50 hazards and which means there's there's a lot of a lot of data that's available there to to utilize. So perfect. Um yeah. So we've talked a little bit about how we each use AI in our daily work. Um and can you say a little bit more about how you individually do that and what are ways that people can can utilize this themselves in appropriate ways? Um and Dan, in particular, how do you coach other practitioners in the areas of crisis communications on how to use AI? Where do they start?

SPEAKER_00

Yeah. So um I advise them on a few things. First and foremost, I suggest they use AI to do everything that's mundane. Uh what AI can do for you is taking care of all those tasks that don't take uh that don't need to take up your time. So why I'm an advocate for it writing the first draft of all your products is because while it's doing that, uh the crisis communicator can be providing leadership counsel, can be telling the story, and could be, in our world of emergency management, literally helping survivors and saving lives. Um, that's the value. That's where that's where the evidence comes in that we have to put our egos aside and embrace the best parts of these tools. There's never been a tool before we're like, I'm not gonna use a rake because I could clean, I could clon the grass off my yard. No, of course you're gonna use the rake. Tools are there to help with the mundane. And so that's the first thing I rec I advise them on. But then it doesn't just end there like, oh, use it and you're fine, right? So I also advise them that the approach is important. I think that if you fear it too much, you err on one side. If you're dependent on it too much, you err on the other side. I think that we should never let artificial intelligence think for us. We should, in fact, often question it. Even in my personal life, when I'm using ChatGPT to work through some concepts and ideas, I've pushed back. And I can tell you, and if you haven't, it for folks listening, if they haven't seen this, they probably haven't been on AI enough. I've had ChatGPT come back to me several times and say, you know what, you're right. Like we disagreed and it realized I was right because it didn't understand all the context I knew. Which leads to my next piece of advice, you only get out of it what you put into it. So again, in our message prison tool, we didn't just say three of us built this thing and so it's cool because of us. We put in every FEMA document, every FEMA SOP. We put in all the information from ready.gov on before, during, and after disasters. And you can't get that in one simple conversation with Chat GTP. Uh lastly, I'd say uh don't ever let your AI have the final word, right? You you you have to have the final word. It doesn't replace us, it's there to assist us. And I liked what Eddie said about the dog but and the leash, but I I if you don't mind, Eddie, I'll adapt that metaphor in a slightly different way. My advice in general that encompasses all of this is don't fear that you're being controlled by the dog or by the AI. Control the dog or the AI, right? We have the leash, we get to decide how it's used, when it's used, and what protections we put in place.

SPEAKER_02

Yeah, and then Jeanette, I would say for mine, I my guidance would be it's not ready out of the box. That that's where I think people get into a lot of trouble. Is people who are like, you know what, I'm gonna adopt this. I see people doing it, let's just do it. And they'll open up one of the generative AI platforms, they'll type in something, it pops up, they're like, okay, now I'm with everybody. I understand. And they use that as a final document. And they don't understand the principles behind it. And so they know enough to be dangerous, and and that can get them into trouble. Um, so for I guess for my area, um, that's that's a big deal. So if this is something you're looking at or anything else, take the time to learn about it. Take the time to reach out and and talk to other people who have been in this for a long time. And going back to your social media example in the beginning, which was fantastic, um, you had those agencies and organizations that were late adopters to social media, and they just tried to go like full throttle and said, okay, you know, we're doing ice bucket challenges for cops, we're doing all these things. They did not have the understanding, and it did not go well at all. Um, and so then they're like, you know what, maybe we need to invest in people who understand this. Um, because one of my favorite things to question when we do training is, you know, um, because we use social media for alerts, warnings for moving information, great tools. I'll be like, well, who's in charge of your social media? And oftentimes it's an officer who'll raise their hand, well, I am, you know, I'm the current PIO, right? Um, just assigned. We're not talking major trained, and right, just a little bit. And so I'm like, okay, so you're in charge of that. Well, how many social media accounts do you have personally? And sometimes I get the none. I don't like it. And yet you're in charge of your agency's social media, or I have one and the list, you know, Facebook. And it's like, okay, you know, Gen X, you know, like like who who are you? But then you get these other people who who really have that in-depth understanding who can provide that guidance. And that's those, or should I say, those are the people we need to look for to help us. And the same thing with social media is with AI technologies. There are people who understand it, who can use it, know how to apply it, and you need to lean on them. Just because you're a higher rank folks does not give you knowledge in this area. And and I get that across multiple agencies, and I'm sure Dan, it goes with, you know, different positions within FEMA and everything else. We look to people of higher rank and we're like, well, how should we use AI? I'll be honest, folks, um, our our baby boomer generation and those that are coming after truly have been through more changes than I think in any other time in the world. And they've they've gone from radio to where we are now, they've dealt with a lot. But just because you are that higher rank, give you know, if you are that higher rank and listening, ask some of the people that are in this space. They're gonna give you, I think, the best guidance on how to move forward.

SPEAKER_01

Yeah, I I think those are both really important um you know recommendations about how to use AI in your daily work and also the cautions about who's doing it. You know, when I when I use AI on a daily basis, I I've just dipped my toe in my um I normally I use just Google for searching for information. And it gives you the option of do you want Gemini to help you? And so I'm just dabbling by saying, Yeah, let's see what Gemini can come up with. But Gemini always provides the reference material for where it's giving you its bulleted list of answers. And sometimes, even there, the answer that it provides sounds correct, but it really does need that extra human check. Because I'm frequently asking things about things that are within my area of research. And I know even though it's gotten the right words from the document, it's not the right words to answer the question that I'm asking. So, you know, even in dabbling in this, I think it's it's this is just one way of getting your toes wet. Um, but it also re-emphasizes to me the need to have really deep understanding of my subject matter so that I can make sure that when the agent returns this information, it's returning something that's accurate, um, which really gets down to the next question about the place of the professional and the importance of, you know, we all three of us have kids who are entering the workplace, they're in college, or we have um students that we work with who are preparing to launch their careers. And honestly, there's a lot of fear about where is their place gonna be in emergency management, in their profession that they've chosen. Um, and AI is kind of that looming ominous specter that it feels like it's taking those entry-level jobs and you know, where are they gonna plug in? And yet when I think about, yes, AI can certainly do these tasks that we've been talking about, and it's really great. But the knowledge base that is a requirement from anyone who's coming into these fields is going to become even more important as that deep check on the information that is being generated by these tools, um, uh regardless of what domain you're working in. Um, so how how should we be looking at AI use for risk and crisis communicators and for writing alerts and warnings for these these next generations that are that are coming?

SPEAKER_02

So I'll throw in first here and say I'm excited for the future. And I'm excited for the people that are just coming on because I think they are uniquely equipped to deal with these types of changes. Um, when you're talking to someone who's been an emergency manager for 35 years, this can be a very tough thing. If you're talking to someone that's currently in college, I mean, Jeanette, you and I have there are a lot of good programs out there that are training um the future, the next generation of uh people to work in this area, and they're talking about AI, they're talking about its applications. Maybe some have dabbled too far academically and have had to learn some tough lessons. Right. Um, but they're prepared. And and you think like auto mechanics have been around for a long time. Um, I mean, there are people that talk about the early cars, they were easy to fix, but cars developed, you know, more technologies. They're computer-based now. I it is hard to just fix them on your own often. And it's not that mechanics went away, it's specialized mechanics became so vital. And it really gave opportunities for those newer mechanics to say, hey, I'm not only bringing the mechanical insight, but I'm understanding technologies. And I think they became even more in demand. And so I think some of the areas that are at risk are the ones that are resistant to change, that have been there for 25 or 30 years that are saying, I don't want to take this next step. They're the ones I think that can be replaced or are gonna have the most difficulty. For everybody who's new coming in, if you're listening, you're in the right place, the right time, learn about this stuff, learn about it's the the right way to apply it, and you're gonna have so many opportunities. But I'm an optimist, Jeanette, and you know that about me. But that's what I that's what I think.

SPEAKER_01

What about you, Dan? I mean, I know you work as a professor also, and you have students that you're talking to as well as mentoring public information officers. And there's, you know, what's your advice to to this generation?

SPEAKER_00

You know, one of the one of the pieces of advice, especially to my students and is uh and and people that I know who are coming in from a negative perspective, is nobody's asking you to devalue humanity. In fact, now's the time to value humanity even more. We should value our unique perspectives, our our our intuition, our special voice that we have. Uh, because by valuing that, we're going to have a better and more effective relationship with the artificial intelligence we use. And I'll give you an example. A couple of colleagues and I, we had a disagreement recently, and one of them fed a problem into Claude, and another fed it into ChatGPT, and they came back and they got different answers. So now we have AI fighting over AI, and I was just laughing at it. And I'm like, the question isn't what did they both give different? The question is, what did you both input different? What was the humanity that you gave, the questions that you gave? And that's where it becomes critical. Our humanity is more important than ever. And so we need to, if we want great results, we have to have great input. And AI can't give us the input to give to AI, if that makes sense. We're making that choice as a human. And and and I would add to that, uh, I also tell them, though, a complete advocate like Eddie, I'm excited like Eddie. This is the greatest thing, and it can be the greatest thing. But we talked a lot about the macro level. On the micro level, we need and have a responsibility to pay attention to the nuances we learn from AI the more we work with it. So, you know, ChatGPT and the rest of these generative AIs, they're taught to be diplomatic and to be nice. If you just ask it, hey, I have this idea, blah, blah, blah, it'll find nice ways to placate you and tell you it's okay and it'll work this way. If you ask it, I have this idea, come up with every way you can tear it apart, you get a second one. If you say, I have this idea, before you give me your opinion, ask me a bunch of tough questions, you get a third result. Um, and then lastly, just because we're both uh uh have college students have a chair, you know, I also like to caution the students, um, there are so many tells. If you know from poker, a tell, we can tell what your hand is uh for using uh AI. Uh so the word choice, AI just loves certain words. It loves certain grammar, it loves the m-dash. If you haven't, you know, heard that. So as a as a writer, if I see one M-dash in a paper, maybe. If I see seven M-dashes, I know that this was written by AI. And that's just one example. There's so many others that I won't go into, but the only way folks will learn the things that you, I, and Eddie have learned is to dig in, bring their humanity to it, and play.

SPEAKER_01

Yeah, thank you. Uh uh, I think they're the the tools are evolving quickly, and we're going to see this great improvement in their capabilities as they continue to evolve. And the personal uh experience and expertise that individuals bring will remain vital to create to giving those tools the human touch that's so necessary. And also for validating that the information that it's creating is correct, accurate, trustworthy, credible. And all those things that you want on any press release, alert and warning, social media post, anything that's going out, that human touch on it before it's issued is so important. Um and will continue to make a difference. So um, but this has been a really fun conversation about AI, and I particularly appreciate it that none of us are really technical. You know, we didn't have a computer scientist on this or somebody who does all of the programming. And so I think that we've been speaking language that is probably common to a lot of our colleagues who are out there trying to understand how do I even think about this as a problem, as a topic space. How do I approach it? Um so um what what are your concluding thoughts as you're thinking about who our listeners are and what they might be going away from this conversation with?

SPEAKER_00

Our our approach to AI is within our control. Our use of AI is within our control. And we can use AI as a tool without ever losing our humanity, our voice, and the difference we can make individually as humans.

SPEAKER_02

Yeah, I'd say if you're one of those that are just naturally resistant, do a self-check and really figure out why that is. Um, because maybe some of your concerns, while you know, could be valid, others um could just be based on things that you've heard or read. Um and AI is not the the end all be all, the one that's gonna save you. Um it is a wonderful tool and it works well until it doesn't. And in that time when it doesn't, um, you need to have the same tools and and everything else. Now, I cannot be as fast as a computer. I there is no way. But when I'm out in the field, if I'm walking through water, if we're doing evacuations, if we're doing something, and I have to get a message out right then, I need to have the tools and experience where I can just open up, even if it's I have to write it down on a piece of paper to get it to somebody, to get it into the system, I need to know how to do that. And so while I can use these tools for so many different things, if I don't have the the knowledge, skills, and abilities to to also support that, then that's where a big failure is. Um so use this stuff, but you need to make sure that you're just as trained um to go forward.

SPEAKER_01

Yeah. I guess my my concluding thought would be um as an academic, I like to see evidence of accuracy. I like to see experiments with outcomes that are being measured. Um I want to see how things are gonna perform based upon the metrics that have been created to say this is what is good and should be relied upon. And so um those benchmarks are coming within at least within the space of alerts and warnings. And we have experts like Dan, who has worked in the public information officer space, crisis communications, and helped to design technologies and applications that can be used there. Um and I think the importance of drawing from these focused AI models that can help you to do a specific job well, um, that is where I would be most inclined to put my greatest trust as we're moving forward. But things are changing.

SPEAKER_02

I was gonna say so, Jeanette, where do you rate yourself going off of your social media example? Um, because folks, if yeah, if you're not already on board, you're you're a late person now into this game. Um where because you said you're making adjustments. And Dan, I'm gonna say you're an early adopter, early, early in. Um where do you rate yourself on your social media scale? If this is AI.

SPEAKER_01

I'm like a resistant early adopter. I'm I'm um I'm learning about it and thinking about how it can be used appropriately, but I'm not yet fully in and I'm not utilizing it routinely. It's like a dip in, see how it does in this space, but then I'm still like a spreadsheet, manually entering data kind of person because that's what my comfort level is still. Um but so I see the value and the benefit and I know it's coming, but I have I'm not at the beginning of the curve of being like an evangelist saying everybody should be adopting. Um so I'm hesitant, but um also believe that the tools are improving and it will be that a point in the future where I will say, yes, I'm on board.

SPEAKER_02

Well, and to be fair, for those that don't know, so Jeanette and I talk about this often. She is very much in the research space of this. And and she is not just on the sidelines, you know, reading a news article about this. She's definitely um neck deep and trying to understand these things. So um that's why I'm asking her where she is with all of this, because you know, I again I am uh I don't know if I was an early adopter, but I definitely have come on board and I'm trying to train and learn. Um, but again, for the work that Dan that you've done since I I would say you are an early adopter, um, I'm hoping that you're seeing people become less resistant as they learn more.

SPEAKER_00

I am. It's just it's a slow process, and that's why uh this panel discussion is just so important. And and I think it needs to be heard far and wide, because I just noticed that not only do we come from three different backgrounds, experiences, and professions, we just realize that we're at three different levels of hesitance, uh, embracing evangelism, you know, early adoption, and and that's okay. Um, but we're all on the path. Right. So uh I I think that's a path that everyone needs to to keep to keep walking down. I do see change every day. It's it's slow. Um, and I think it's worth speeding up a little bit and catching up with the tools um so we use them the best way possible.

SPEAKER_01

Well, thank you. Thanks, thanks for for joining our panel discussion on AI, crisis communication, alerts and warnings. I think this has been really fun for our listeners. If this is like the kind of format that you want to hear more about um, if you want more panels, let us know. Because it's I think it's fun to have these varied perspectives as we each come from a different discipline and different ways of thinking about our tools and technologies. So um thanks for being part of this today. And Eddie, do you want to close us out?

SPEAKER_02

Sure. You do so good too. Folks, this is the Alerting Authority. Just as she said, let us know your thoughts and opinions. We are here for you. Um, we love having these types of interactions, and we want you to feel supported every step of the way. Um, we understand that every second has a story, and we're grateful to be here to help share yours. Have a good day.