Episode 235: Ethical Considerations in AI Implementation for the Fire Service with Scott Roseberry & Sai Narain

Episode notes

What's the future of AI in the fire service?

Hello Smart Firefighting Community!

Welcome to another episode of covering real world innovations via interviews with fire service and technology industry experts that empower YOU to develop your very own Smart Firefighting strategy!

In this episode:

How is AI transforming fire service operations?

How does AI enhance EMS for better patient care and safety?

How can AI act as a virtual assistant for first responders?

Hear from Scott Roseberry - the Battalion Chief at Garland Fire Department - and Sai Narain - the Director of Technology Strategy; State & Local Government at Microsoft.

The two talk about the application of artificial intelligence (AI) in the fire service and public safety. They explain that AI is like a computer brain that can learn, adapt, and assist in decision-making. They discuss various use cases of AI in the fire service, such as improving communications, incident command, and community risk reduction.

Scott and Sai also touch on the ethical considerations of using AI and the importance of responsible AI implementation. They envision a future where AI acts as an extra person in the fire service, helping with decision-making and improving firefighter safety.

So take a listen to hear what they have to share!

Head to www.smartfirefighting.com to discover how SFF accelerates innovation for emergency responders, to find out when our next event is, or review our curated resources!

Facebook | Instagram | Twitter | LinkedIn

Chapters

1. 0:00 Kevin's intro: AI in the fire service
2. 0:55 Scott Roseberry and Sai Narain join
3. 1:57 What AI is, in simple terms
4. 4:10 AI for communications
5. 8:38 Obstacles to adopting AI
6. 12:27 Responsible AI and ethics
7. 15:43 AI is a tool: learn by doing
8. 19:39 AI as a game changer with more data
9. 24:03 AI for wildfire detection and the wildland interface
10. 27:15 AI in EMS and on the ambulance
11. 29:58 AI as an extra person in the jump seat
12. 33:57 How to get in touch and get educated
13. 36:02 Wrap
14. 36:40 Outro

Questions answered in this episode

What is artificial intelligence in simple terms?

Sai Narain describes AI as a computer brain that can learn, adapt, and assist in decision-making. Examples include pattern recognition, making predictions based on existing data, filtering spam emails, music recommendations, and voice assistants on phones.

Hear it at 2:47

How can AI help with emergency radio communications?

Scott Roseberry explains that AI can listen to radio communications in the background and recognize certain warning terms. He references Don Abbott's Project Mayday research, which identified 16 terms commonly said before a Mayday. If AI hears two or three of those terms, it can warn the incident commander that a Mayday might be imminent.

Hear it at 6:46

What does responsible AI mean in the fire service?

Sai Narain and Scott Roseberry emphasize looking at AI as a co-pilot, not a decision-maker. AI should surface information to first responders, who then make the final decision. Ethical use includes fairness, transparency, accountability, privacy, and reliability, with the human retaining responsibility.

Hear it at 14:10

How can firefighters start using AI today?

Scott Roseberry suggests getting a free account with a large language model like ChatGPT or Microsoft Copilot and experimenting. He says AI can help write or translate public service announcements, draft SOPs and guidelines, and analyze data from Excel spreadsheets. His department's battalion chiefs now have ChatGPT accounts because of how much they use it.

Hear it at 16:38

What role will AI play in wildfire detection and response?

Scott Roseberry says AI is being trained to watch thousands of cameras across the country 24-7 to identify smoke and distinguish it from mist or fog. Unmanned aerial systems are being developed to operate fully automated, and AI can model high-hazard areas, fire behavior, and better evacuation routes.

Hear it at 25:05

Transcript

Machine transcript of the published audio. Names may be misspelled and some sections are missing.

Ep 235 transcript

Ethical Considerations in AI Implementation for the Fire Service with Scott Roseberry & Sai Narain

Speakers: Kevin Sofen, Sai Narain, Scott Roseberry

Source: Whisper medium.en machine transcription of the published Podbean audio (no original transcript found in Drive), lightly cleaned. Speakers were separated by voice clustering and then checked and assigned by hand at every speaker change. Gaps where the model returned no usable text are marked [inaudible m:ss].

[0:00] Kevin Sofen: Welcome to the Smart Firefighting Podcast. Kevin Sofen here, where we cover all things related to the future of smart technology for first responders around the world. Please enjoy the show and let me know what you want to learn about next. In this episode, Scott Roseberry and Sai Narain dive into the transformative role of artificial intelligence, AI, in the fire service and public safety. They explain how AI functions as a computer brain capable of learning, adapting, and aiding in decision-making. We'll explore real-world applications of AI, including enhancements in communications, incident command, and community risk reduction. Additionally, we'll discuss the ethical considerations and the importance of responsible AI implementation. Join us as we envision the future where AI acts as an invaluable team member, enhancing decision-making and improving firefighter safety. Enjoy. Hello, everyone. Welcome back to the Smart Firefighting podcast. Really excited today to be with Scott Roseberry and Sai Narain. Scott and Sai, happy summertime. How you guys both doing?

[1:04] Scott Roseberry: Doing good. Thanks for having us.

[1:06] Sai Narain: Likewise. Thank you for having us, Kevin. It's just getting started. Summer is getting started.

[1:10] Kevin Sofen: I know. It's a wonderful thing. It's sunny, blue skies in Chicago, and life is good, street festivals going on. But here we are. What a surprise. The three of us have a little round table talking shop on all things fire tech, particularly artificial intelligence. It's one thing that's been in the news in more ways than one, sometimes better lights, sometimes worse lights about what it is and the concerns. But one of the biggest things is that there's a lot of maybe misunderstandings about what it is, but particularly how it applies to public safety. And there's an exciting article that we're going to dive in here today that you both have had a hand in. And I think in the simplest terms, I'd love to hear from both of you who have been practitioners in this. Artificial intelligence, what is it and how does it relate and how does it apply to public I'll throw that question to you first, Scott.

[1:57] Scott Roseberry: Well, I'll let Sai kind of do the definitions of the different types of artificial intelligence. I think he'll do a lot more justice. But that's kind of how does it apply to the fire service? And that's really what got Sai and I kind of talking about this. We've collaborated on a few different things and we just got talking one day about how is AI going to improve or affect the fire service? And I just started kind of throwing problems out there or issues we have in the fire service. And then he just kind of was talking about, well, here's what it can do in these different instances. And so we decided, we wrote an article together, co-authored an article for Fire Engineering Magazine, based in talking about the different points of whether for communications, wildland urban interface, instant command, just different aspects like that. And then what's the issue today and what will it look like with AI in the future?

[2:47] Sai Narain: True that, and I think in simplistic terms, AI essentially is just like a computer brain that can learn and adapt and assist in decision-making. You know, whether it be a simple pattern recognition or making predictions based on existing data, you know, we're all using AI on a day-to-day basis, whether it be, you know, listening to our favorite songs on a music platform that we all love, or whether it be, you know, filtering our spam emails, even, you know, giving driving directions or using your phone on a day-to-day basis on a voice assistant. I mean, AI is sort of available on a day-to-day basis across the board. And it's just sometimes we don't realize that we're interacting with AI. And AI is many forms and shapes. You can take everything from like narrow AI, which is essentially just doing one specific task, to generative AI, which is our new flavor of AI that everyone's been talking about for the last year. and which has really created this big buzz of conversation and then everything in the middle, everything from machine learning, you've got deep learning, you've got natural language processing, image recognition, et cetera, et cetera. So AI is a broad term, been around since I would say, gosh, back in the 50s, taken various shapes and variations, but it's definitely gained a lot of attention over the last year.

[4:10] Kevin Sofen: Thank you for that. And I think just now I want to continue to dive into this sort of translation for the fire service of sort of what it means. The fire service is dealing with a range of inherent challenges, whether it's with communications, whether it's with the violent fire events, whether it's with incident command, whether it's with the rise of wooey fire environments, there's all these challenges. Fire service is now no stranger to challenges in overcoming these obstacles. But I'd love if you could maybe paint a light on some of the one or two of these use cases, maybe let's start with communications and incident command about kind of explaining a little bit of the chaotic space and the limited cognitive bandwidth and operating and stress environments and then sort of this new component of artificial intelligence as one of those, seven types of AI that you mentioned, Sai. What does it look like? How does AI like, again, like how you mentioned like for a music platform where we're experiencing it and like we don't really think about it's happening, but we are experiencing it. Can you paint the picture in terms of an example from the incident command or communication standpoint of how AI sort of plugs into the experience for a first responder?

[5:23] Scott Roseberry: Yeah, so on communications aspect, we can break it down into emergency and non-emergency communications. Non-emergency, I mean, our main form of communications is email. Like Sai just kind of touched on, AI is already scanning your email. So you may notice that your spam folder, more and more emails are getting diverted to your spam folder these days. That's because AI is scanning those, every email that comes across. But I think we all fall into the problem of emails getting buried. We get so many emails, the important ones get buried, and how do we keep track of that? And future AI could help us keep track of that. It's gonna notice, it will notice which emails need to be responded to and haven't been responded to. Some systems are already doing that, but I think it'll be even better in the future and remind us which emails we really need to do. On the mercy ground form of it, I go back to Don Abbott's, all the work he did on Project Mayday, and one of the things he came up with were 16 terms that are said before Mayday happens, and they're pretty common. And he did thousands of research on thousands of departments in Mayday calls. And one of the things AI can do in the background, the sign I talked about, was it can listen to our radio communications, and if it hears, say, two or three of those terms in that Project Mayday, it can warn the incident commander, hey, you might want to start thinking about pulling people out because May Day could potentially happen here. So AI can definitely help and listen to the background of our communications.

[6:46] Sai Narain: Yeah, I think you brought up a really good point, right? It's filtering information from the noise because over the years, the data overload continues to rise, right? Like, leave alone firefighting, like us in our personal lives on a day-to-day basis, from the time we wake up till the time we go to bed, the number of notifications we get on our phones has just astronomically increased and with notifications don't just come out of the blue, right? They come because of data behind the scenes. And so we as humans are ingesting so much of data on a day-to-day basis. We're not ingesting really information at that point. I see the biggest value of AI is in separating that information from the noise and helping understand what is important and what's not. The examples that Scott was sharing, non-emergency communication, I And I give an example, there are many cities and counties across the country or across the world that 911 call takers take non-emergency calls when they're not on a 911 call, just because the non-emergency calls are also routed to the same dispatchers. But if you're looking at a busy dispatch center that's responding to a lot of 911 calls, a lot of the non-emergency folks are sitting on the phone for a couple of hours sometimes. And so there are many ways in which AI can help in these non-emergency scenarios as well, whether it be simply providing information based on an existing knowledge base, contact the permitting department or here's the address for the space or whatever, right? And then when you start looking at the emergency situations, just like Scott talked about, separating that information from the noise, looking at keywords, looking at patterns, And again, from the lens of a co-pilot, what that essentially means is you're providing that, surfacing that information up to the first responder and letting the first responder make the determination or the decision on what to do next.

[8:38] Kevin Sofen: Right on. And I feel like there's a big obstacle that we face in terms of really seeing that and experiencing until you sort of maybe see and experience it firsthand. I mean, there's, as we know, there's that terrible phrase in the fire service to change and reluctant to change over time, and I wouldn't even say it here on this podcast, but I think we know that there is some of this challenge of implementing new stuff and a little bit sometimes reluctance to change. From an accepting change and a change management perspective, what have you seen in terms of any early departments that have been like, hey, here is a roadmap on how to do it, then here's how we won over different stakeholders, whether it's all the way from the firefighter the fire chief to city council to the city itself kind of worried about whether it's everyone quickly throws up cybersecurity concerns or privacy concerns, things like that. The overall arching challenge of the change and the change management process. Can you both speak on that?

[9:37] Scott Roseberry: I think everybody's still trying to figure it out right now. It's such a new landscape and nobody's developed a roadmap for us. So every department right now is just trying to figure it out. And there's a lot of things you can do right now to use AI in your everyday operations. I know a lot of us have had that question like, this is great, but how does this help us now? You know, Sai and I were kind of talking about the things that you... that you can do today with it. One is like public service announcements. You know, AI can help you with public service announcements and help you develop those. It can help translate those into different languages than just your, the chat bots. I've been playing around with those lately too, where you can put all of your SOPs and guidelines and protocols and everything you want to in the chat bot. And then it ends up being basically a frequently asked questions kind of thing for your fire department. But there's several areas right now, just helping to write your SOPs and guidelines. And the great thing about it too, is you can put one of your current ones in there and say, hey, can you write it in this form or this template? And can you have an SOP for our response to hazmat incidents? And of course, the more information you give it, the better it comes out. But we've been doing a few guidelines and procedures in our department right now that way. In fact, all the operations battalion chiefs have a account for CHAT GPT through the city because of how much we're starting to use it. But yeah, like I said, nobody really has the remote roadmap right now. We're all trying to figure it out because of how new this is.

[11:06] Sai Narain: And, you know, to your point, Scott, like I think there is two aspects to it. There's the end user aspect to the change, right? Which essentially is just, you know, us integrating AI into our lives on a day-to-day basis in our work. And then there's the aspect of, well, Is AI providing a fair response? Is it reliable? Is the information something I can look at and is it making the right determination? Things like that. And so it is really important to understand the framework of using AI in a responsible fashion, understanding what safety and reliability means, understanding are they set up in a secure manner? Are you putting your information in a place where it's, you know, set up the right way, the right people have access to it. And engaging with AI, you know, transparently, you know, are these systems understandable, the ones that you're interacting with. And so, you know, making sure that you're sort of grounding yourself in a responsible use of AI is so important. There is a learning curve with any technology and AI is no different. There is to be a learning curve. And when I say learning curve, not just on the end user behavior, but also in understanding how this layers into the rest of your day-to-day operations. It's so valuable to make sure you're grounding yourself in that exercise and going deep.

[12:27] Kevin Sofen: And you both kind of mentioned the word responsible AI and then the term ethics was popping into my head. It seems like that's a kind of a gray area now, but particularly when thinking about if we're using AI to help make a decision and something goes great, okay, awesome, but something goes wrong. Again, I know when certain things happen in public safety, really in any industry, it's always quicker through the book in terms of pointing the finger who went wrong and things going to court and legal issues. And I know this is maybe a slippery slope of a conversation, but very important when the biggest challenges is the ethics part of it and sort of, if we're using something made by AI to help us make a decision that leads to a, again, a great outcome or a terrible outcome, pointing fingers at who's responsible. And then I know there's other ethical and responsibility questions, but Can you both elaborate a bit more on the ethical responsibilities side of it?

[13:19] Scott Roseberry: Yeah, so I think one of the areas you're going to really see a lot of advancements in AI and coming soon, in fact, there's already a department working with AI is in the EMS area. And I think that if you think about how AI can help you with identifying patient conditions based on history and current signs and symptoms, drug interactions, you know, what drugs should you give them and what is that going to interact with the other drugs that they already have? So I think that you're gonna see a lot of advancements in that. But ethically, though, the question is, at what point do you let AI decide what treatment you're going to take with the patient, and especially when outcomes could be catastrophic if you take the wrong treatment? And we all know right now you hear about AI with the hallucinations that it has, where it can kind of make things up and it's not always accurate. So at what point do we let it have that decision-making capability?

[14:10] Sai Narain: Agreed. And I think it's a lot of work that needs to happen, right? And I look at ethics from the lens of fairness. Is it unbiased, the responses, or transparency in the sense that is there a clear delineation between what the AI is recommending versus what the human is recommending? Essentially, some sort of an accountability check, whether it be in the decision-making process or it be in the use of the technology. Obviously, privacy and misuse of data is a key factor of ethics. And lastly, safely providing a reliable outcome, which is why it's so important to look at AI as a co-pilot and not really as the decision-making entity. It's like having another set of eyes. It's like having that invisible or the virtual agent by your side who is essentially just assisting and providing that information like we talked about earlier and not necessarily saying, okay, AI said it, so I'm gonna do it. That's not what we're talking about here, right? So again, there is a curve and as technology continues to mature, as governance and policies continue to mature across the world, and as more and more people, you know, build guardrails around these technologies, things are gonna continue to change. And so looking at AI as a co-pilot, if you will, allows us to operate a little more with a sense of comfort that, okay, I am using this, but then I'm making my own decisions.

[15:43] Kevin Sofen: Yeah, it's a tool. We still need to use this little tool on our own in between our two ears. Our brain that we have, I personally look at and utilize a range of AI tools as you stay, ultimately get from zero to one. I still need to go from one to a hundred, but it's a lot of times a starting point, which I think is a good place for people to think about AI in terms of, again, just starting to play with tools. And if you were gonna speak to someone that's like, maybe on the fence or thinking about trying to tinker with various tools, and again, maybe they already are with some of the different technologies and vendors out there, what would you say to a fire department or a first responder that's sort of like, huh, like, okay, I'm kind of interested, but still need to kind of do my own homework. How do you start? You know, where do you start? And maybe any suggestions on a tangible crawl, walk-run path to start implementing some of these tools on departments around the country.

[16:38] Scott Roseberry: Man, I've watched a lot of videos. YouTube's, you know, it's not only great for fake for repairing your car, but it's, there's been a lot, there's a lot of videos out there's a lot of literature out there on kind of how to do, how to get into it, and just kind of baby steps, kind of that crawl, walk, run. So, I mean, get a free account with one of the large language models. There's several of them out there now. ChatGPT is not the only one out there. Microsoft has a co-pilot. There's several others out there that you can get into and just kind of just start playing around with it. See what you can do with it. Start asking it questions and see that the power of it that it has, but you'll also see the limitations that it still has too.

[17:15] Sai Narain: Yeah, and I think, you know, I believe always in this act of learn by doing, right? I mean, play around with the tools, try to understand limitations, to understand like where it works well for you and where it doesn't. You know, it's an opportunity to engage and interact. And I look at it as getting to know each other, get to know your AIs that you're working with, because you can't like it or trust it unless you get to know it. I mean, the same logic that holds good for two people who are meeting, it's like the know, like, trust logic, like holds good with technology and tools as well, is interact with it and see how it goes. The other area I was gonna mention, Scott, is, you know, there are a lot of tools out there that are available today that interact with your day-to-day administrative tasks, whether it be your email, whether it be your collaboration tools, whether it be you're preparing a document, reviewing a document and preparing a presentation, things like that. There are AI tools out there available today that you can start working with, can assist you in those areas. And that, the amount of time that you get back, honestly, the amount of time I get back on my day-to-day basis, just using and leveraging AI in my administrative work, it's a massive opportunity for me to take that time and engage with other things that I've been wanting to do that I've pushed off. I read this funny article somewhere, it was last week, that said, have AI do laundry and dishwashing so I can do the rest. Well, believe it or not, there is AI already in laundry and dishwashing tools that are out there, right? I'm sure there's washing machines that do things a certain way and so do dishwashers. And so the point I'm trying to make is like AI is out there and unless you're putting your hands on it, leveraging it, touching and feeling and interacting with it, you're never going to get to the core sort of operational side of AI in using it for work, unless you get to feel comfortable with the other aspect of it.

[19:15] Scott Roseberry: That's a great point, Simon. Kevin, one of the things we've talked about before is being data-informed. When you're making your decisions, be data-informed. And AI actually right now can do a lot of data analysis for you. All you have to really do is drop the Excel spreadsheet, or it takes several different file forms, drop it in there and ask it to analyze it can do pie graphs, chart graphs. It can do a lot of data analytics for you right now.

[19:39] Kevin Sofen: Right on. And yeah, I think that's, again, kind of looking at how AI can help you become data-informed actually one area of that kind of data form that I personally am intrigued by and one of the sometimes overlooked aspects of the FHIR service is community risk reduction and sometimes doing predictive modeling to kind of, you know, a sort of look at. how can we better prepare the community, but then also then once we prepare the community, what are a bunch of little things we can do to make the community safer, whether it's through home hardening in wildfire areas, or whether it's through smoke alarm installation at homes. But the idea of how can we run various models to help us give ideas, to then implement tactics to prevent the disaster. I heard Dan Muncie one time say, the best disaster is the one that doesn't happen, and so a lot through predictive and community risk reduction. What are your thoughts in terms of some of the predictive analytics to figure out resource allocation to enhance CRR efforts? We'd love to hear your guys' thoughts on that.

[20:37] Scott Roseberry: I think it's going to be a game changer for us, especially as we get more advanced in the future and it gets more accurate. It's going to be a huge game changer. For one, the public service announcements you can do with it right now. I was just doing a presentation for the Texas Fire Chiefs and I was using my phone and I to do public service announcement on reminding drivers not to drive through high water crossings and the dangers of it. And it, I mean, within seconds spits out this great public service announcement on that. And then said, hey, we've got a large Spanish speaking population in our city. Can you translate that in Spanish? And automatically translates it in Spanish. Just that translation point for community risk reduction to get all the members of your community to get that message out to all of them is huge. It's very reliable on those aspects.

[21:25] Sai Narain: Yeah, and I think, you know, I feel like there's a lot more data available, Scott, like today, like that, and continuously, like more data is becoming available. So I think one of the areas I think, Kevin, we can prep is in ensuring that you're starting to collate some of these data sets, right? So traditionally, a predictive analytics has been around for a long time. And what I mean by that is, you know, departments, fire departments do standards of cover assessments or response time analysis and things like that over the years. But a lot of that data is siloed to the department. And I feel like with more and more data becoming available, I know there's a number of efforts nationally and internationally around collating data sets together and collecting data together. I know the Western Fire Chiefs Association is working on some efforts there. And so is like the federal government and many other entities. And so understanding what data is available and accessible, pulling data together, pulling the relevant data together, and ensuring that a lot of the AI is using the right kind of data, it goes a long way. So in a sense, you may not be using AI operationally today in some of these scenarios, but it's important to prepare your data so that AI can assist you in the future. And so what can you be doing now is yes, understanding and familiarizing yourself with AI tools, but also understanding and familiarizing yourself with the data that is available to you and pulling that together, participating in some of these efforts to ensure that you're feeding the right kind of data into it so that in return, as AI becomes more and more accurate to Scott's point or the availability becomes more free to everyone to use, you have fed the right kind of data to those AI tools so that the AI tools are giving you that return on investment, if you will.

[23:19] Scott Roseberry: I think that's a great point too, Sai. And I know Sai's working on an initiative to help with that data points and how you turn those into meaningful actions, but that we as fire service and in the city yourself, your jurisdiction, there's a lot of data points that you're collecting, but you're not utilizing. And AI is gonna help us be able to make something out of that. And for like community risk reduction between your police department, your emergency management department, your building department, all these, your health department, There's lots of data sets in there that are all siloed. And with AI, we're gonna be able to put those together and make meaningful efforts out of it that will help with not just community risk reduction or mitigation efforts in the fire service.

[24:03] Kevin Sofen: Right on, and I know one of the things you talked to both of you in different ways is like the different types of data that's becoming available. And one that I'm personally very intrigued by is some of these passive cameras and sensors in the wildland areas. seeing a lot of movement to having pre-position data creators that are looking for stuff. And then I know there's all types of algorithms built into those cameras and sensors. And then it's sort of creating these different flags of warning or of interest to say, hey, this is something that we need to take action on. But to that point, let's kind of dive into the wildfire wooing environment. Can you speak on in terms of how we're creating new data in different ways, and then how we're using these different types of AI tools to take that data and allow us to take some proactive decision in terms of early detection to early suppression to more effective use of resources and these sort of kind of guiding the wildfire. Yeah, what are your guys' thoughts on that?

[25:05] Scott Roseberry: I think one of the most fascinating things in the wildland interface right now is that, you know, we can put cameras, we put thousands of cameras up all over the country, But logistically, how do you get human beings to watch those cameras 24 hours a day, seven days a week? Or they're training AI right now to be able to watch it. And AI can watch thousands of cameras all over the country 24-7 and identify smoke. And they're teaching it to identify smoke over mist or over fog and identify that for early warning signs. Then you have the unmanned aerial systems, the UAS systems that what they're training right now is to be fully automated. So we don't actually have to have a human flying them that you just give them a you give them an objective and they do it in the least objective for you. And then on the modeling aspect of identifying, you know, high hazard areas, what's the modeling gonna look like that the fire comes, evacuation zones. I was just watching a special last night and they were showing, you know, how the gridlock of people stuck in these wild land fires and they can't get out. So quicker warnings of the evacuate need for evacuations and better evacuation routes.

[26:14] Sai Narain: Yeah. And I think even like a lot of the sensors, like you talked about, there's a unattended sensors that are starting to come up in the field. Like when I say unattended, and I mean, like things like looking at soil moisture, looking at environmental conditions, looking at you know, what these are sensors that are that are either solar powered or otherwise that are just sitting out there and collecting information and then transmitting those bits and bytes into a centralized location, I feel like there's a lot of data coming through there, again, right? And there is a huge opportunity to continue to grab pieces of information to start becoming more proactive, even planning like what areas of vegetation need to be managed in what fashion, things like that. And so while the operational side of the fire service will never change the operations of it, the enhanced situational awareness and the additional capabilities that, the pieces of information that can be brought into that core operational side of the fire service. The value is just massive.

[27:15] Kevin Sofen: And I know one thing that we've learned that spending time together in the TSI and in a lot of the IFC events is the level of EMS involved with all things, public safety and first response. I know Brian Norris down San Antonio has done some pretty cool stuff doing that live video streaming to be able to, when the call comes in, that whole phrase of a picture's worth a thousand words, a video's worth a million, and really cool to see sort of the integration of technology, and I'm not sure necessarily what aspects of AI is involved or not involved in that, but considering how much public safety, firefighters in particular, are leaning into EMS is something that has to be part of this conversation. What are your thoughts, or what's kind of, what's been happening from the integration of AI I would think EMS and how it's sort of same conversation and helping with better, faster, more efficient decisions, helping firefighters and EMTs leverage AI for good.

[28:10] Scott Roseberry: You think about all the technology on an ambulance these days between your EKGs and your different types of tech that you're using and all that can be integrated with AI and help with real-time decision-making with patient care between specific treatment protocols, medication dosages, And just what procedures take based on the patient's condition, you know, based on your established guidelines and AI is going to help improve all of that.

[28:35] Sai Narain: Yeah, I mean, like everything from like rapid assessments, like triaging, I know there's some technologies that are being used in that space today, but I feel like there's there's more obviously predictive deployment of assets, which we've talked about earlier. I also look at connected devices, again, even in EMS, right, picking up information from the connected devices that are being used. Again, some of the data sets are siloed today. They don't feed across the board and create a feedback loop with, you know, a typically, traditionally you've got dispatch, you've got response, and then you've got post response. Like once the patient's transported, you've got the rest of the workflow. These three areas in many cases don't have data that cuts across, but there's massive value in creating that feedback loop between these three. And then also like there are tools that are in use today for patient interaction, whether it be like transcribing a conversation again and preparing a healthcare record report out of the conversation so that the patient interactions are more personal as opposed to someone looking at a device all the time and entering information and removing that personal ability to interact. So again, I feel like opportunities for AI in the future is to make people more personal so that the AI can do some of the things that remove the personal interactions or those aspects of human connection.

[29:58] Kevin Sofen: Love that. Yeah, and you mentioned earlier the washing and drying perspective of what are the different things in the background that AI can do to allow you, human, to focus on the most important task at hand, and to really make sure that you're utilizing your skill sets to the fullest. You guys have covered quite a bit, and there's more, frankly, from just even the different types of AI to all these use cases, they all deserve a double or triple click. I think kind of maybe final question I would love to throw to both of you, and you know, it's June 2024. The future vision for AI and the fire service, you know, we're already seeing some use cases, examples of what exists today, and some early adopters, and some examples of that are really changing the fire service for good. But one, five, 10 years from now, where do you hope the public safety sector, firefighters in particular, are in regards to their relationship with AI and how AI is ultimately going to be a force for good to help make first responders safe for the community, safer, your thoughts and the vision.

[31:00] Scott Roseberry: I think it'll be like having an extra person in the jump seat of the engine. For those of us that ride three people on an engine or truck, it'd be like having that fourth person there to help with your decision making and help with size, talking about do that, the ancillary stuff on the background to give you more time to be able to interact Miss Smith, you know, and get better patient care and treatment to the citizens. It's going to be a disruptive technology. It's going to take a change management style of leadership to start implementing AI into the fire service. You know, we're arguably one of the slowest sectors in the public safety or public sector to adopt technology. And it's going to take, you know, change management from the leadership to help implement technology. We have a lot of firefighters that are resistant to it I think it has great potential to improve our job and supplement our job and not replace our job. And I know that there's probably a fear of that it's going to replace firefighters. And I don't see that happening in the near time. There is some amazing work being done with robotics. And when you couple that with the ability of artificial intelligence, could robots one day be going into house fires? It's possible. You know, the interior robotic fire fighting has actually already happened. It happened in 2019 to Notre Dame Cathedral fire. I used an interior robot to do interior operations to help protect the cathedral. So that's already happened, but I don't see it replacing us completely in the near future.

[32:31] Sai Narain: Yeah, and I agree with 100% with what Scott said, right? Like, just because there's another person in that vehicle responding to that incident, in this In this case, AI doesn't mean, like even today, if you think about it, like you add another, you know, EMT or paramedic to an incident, it doesn't make all four or three people right. And what I mean by that is it doesn't mean like all, everybody's going to agree on everything that everyone talks about, right? There's a collective assessment and there is a collective response. AI as part of that collective response. The idea that, yes, it's another virtual agent in that team, and then there still needs to be a collective response. There still needs to be a collective agreement to move forward. And so I would just say, I would love to see, you know, AI play a huge role in firefighter safety as well. You know, I think that there's some massive opportunities in that area, you know, especially because as more data becomes available, They're similar to, I know the other area that we talked about was a proactive sort of community risk reduction, Kevin. I feel like there's a massive opportunity in that space as well. There's so much to do, but you can't get there unless you get started. So I would just urge folks to say, okay, get comfortable with it, interact with it, get to know the unknowns so that you're not faced with this steep learning curve because the rest of the world is gonna continue to adopt AI in different shapes and forms.

[33:57] Kevin Sofen: Well said, and yeah, so I guess I ask you guys, both the future vision and in great context there, but I think to maybe have one final sort of call to action or ways to collaborate with you both or ways to get involved, any sort of final thoughts or could you both maybe give any indication on if for public safety members or startups or people were say, hey, I was interested in getting involved with you, any sort of final thoughts and ways for people listening today to get in touch?

[34:26] Scott Roseberry: Definitely through the IAFC technology council, the vice chair of that, you can definitely reach out to us through that. And when you publish this, I mean, you're free to put my contact information, my email on there. So free to email me in with any questions. Like I said, I don't have the roadmap, but I'm trying to build one. And I'm out there trying to see what we can do to use artificial intelligence to really improve our ability to protect and respond to the citizens needs.

[34:53] Sai Narain: Likewise, Kevin, I mean, feel free to put my info out there, but I think, you know, the IAFC Tech Council is doing some great work. The Technology Summit, I know, is coming up again later this year. I think it's a great opportunity to connect and interact and understand how technology interleaves with operations, and I think that's also a great opportunity to interact with some individuals who are pushing the envelope in that space. And then the only other thing I would say is, you know, get educated, like learn, right? I mean, take a little bit of your time and learn about it. Like, look, I'm a technologist and I work in tech strategy, but I can tell you there's very, there's so much that I don't know, so much that I have yet to learn, and this thing changes every day. And so it's perfectly okay to not know, but it's equally important to maybe just learn a little bit more, right? And so for me, the way I keep in touch is I just learn a little bit every other day. I just kind of pick up an article, pick up a LinkedIn learning or something and watch a video. So just get educated in any way you can, because that's the only way to stay anywhere close to the current.

[36:02] Kevin Sofen: Yeah. Well said, and part of it is staying in touch with guys like you, Scott and Sai, so I appreciate you both being practitioners of this sort of blue ocean opportunity and then figuring it out and helping the public safety sector and first responders as a whole. So excited to see the article come out. Excited to see both at FRI, I'm sure, and TSI later this year and just to keep building momentum in the tech council and all things public safety and using tech and AI for good. So Scott and Sai, thank you both so much. Really enjoyed getting to dive deep here today and look forward to crossing paths again soon.

[36:34] Scott Roseberry: Thank you, Kevin. I appreciate you taking the time.

[36:37] Sai Narain: Thank you, Kevin. Really appreciate it. And Scott, as always, great chatting.

[36:40] Kevin Sofen: Thanks for listening to today's episode. What did you think? What'd you learn? Hopefully something, right? Please reach out on social media with your thoughts on this episode and any other suggestions for future content. We look forward to capturing more important stories across public safety and are here at your service.