--- id: "workshop:from-plans-to-preparedness-using-gen-ai-as-a-practice-partner-in-the-public-sector" type: "InnovateUS workshop" title: "From Plans to Preparedness: Using GenAI as a Practice Partner in the Public Sector" url: "https://innovate-us.org/from-plans-to-preparedness-using-gen-ai-as-a-practice-partner-in-the-public-sector" date: "2025-02-19" duration: "60 minutes" language: "English" instructors: ["Michael Baskin"] recording: "https://www.youtube.com/watch?v=AvbsISUAIbU" transcript: "automatic, from the recording" source: "InnovateUS (innovate-us.org)" licence: "CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org" --- # From Plans to Preparedness: Using GenAI as a Practice Partner in the Public Sector InnovateUS workshop · 19 February 2025 · 60 minutes The wildfires are burning—how do we respond? Cities have plans, but are they truly prepared? In a world of rapid change and emerging challenges, public sector organizations need more than plans—they need readiness.  This workshop introduces GenAI as a “practice partner” for governments, offering a low-cost, high-impact way to prepare for extreme weather events, misinformation campaigns, and other complex emergencies. Participants will explore how GenAI can be used to run dynamic scenario exercises, tabletop simulations, and role-playing activities that adapt in real-time. These exercises help teams build agility, improve collaboration, and strengthen their capacity to respond to the unexpected. Drawing from real-world examples, including a recent wildfire scenario run for city managers, this session will show how GenAI can help public sector leaders practice and refine their responses, becoming more adaptive and prepared for emergent futures. ## Learning goals - Understand the role of GenAI as a “practice partner” for dynamic preparedness exercises - Learn how to create and run scenario-based training for city executives  and teams - Explore real-life applications of GenAI in disaster response and emergency planning - Gain practical tools for integrating GenAI into preparedness efforts to build more agile and adaptive organizations ## Speakers - **Michael Baskin**, Chief Innovation Officer, Montgomery County, Maryland Recording: https://www.youtube.com/watch?v=AvbsISUAIbU ## Transcript Automatic transcript of the recording: names and terms can be misheard, and the times can be a few seconds off. Each time links to that moment of the video. The transcript alone: https://innovate-us.org/from-plans-to-preparedness-using-gen-ai-as-a-practice-partner-in-the-public-sector.txt (plain text) and https://innovate-us.org/from-plans-to-preparedness-using-gen-ai-as-a-practice-partner-in-the-public-sector.json (every segment with its start and end second). **Michael Baskin** [0:00](https://youtu.be/AvbsISUAIbU?t=0) These rooms, let's make them even one step bigger because we keep having folks join, which is awesome. Wow, lots of folks. Okay, great. So these two questions describe your experience with practice that's gotten you ready and share time when you weren't ready. And if you're just joining, the first set of breakout rooms was where do you come from today and how are you currently using generative AI? All right, we'll come back together in four minutes. Here we go. All right. [1:23](https://youtu.be/AvbsISUAIbU?t=83) Leslie, Mary, Pablo, Bill, Chris, Jenny, we are having an interactive session. Susan, Tucker, welcome. Folks are in breakout rooms, so I'm going to add you into those existing breakout rooms. **Audience** [1:38](https://youtu.be/AvbsISUAIbU?t=98) Hey, Michael. Can you do us a favor? The AI chatbot that's floating around is prohibited in some of my team member states in our room. So if you could kick out the AI because I can't, that would be great. And then you can kick us back to our room. But we had to come out for a **Michael Baskin** [1:54](https://youtu.be/AvbsISUAIbU?t=114) second. Okay, got it. Thanks, Maggie. Thank you, Derek. You're welcome. Participants. **Audience** [2:03](https://youtu.be/AvbsISUAIbU?t=123) Let's just boot these people. Would you mind booting the folks that are AI? **Michael Baskin** [2:10](https://youtu.be/AvbsISUAIbU?t=130) Yes, I'm taking care of that. Super duper. Thank you. Awesome. And folks who have just joined, I'm going to add you in. Hey, Anniket, nice to see you into these groups. So you can join on it. **Unknown speaker** [2:29](https://youtu.be/AvbsISUAIbU?t=149) I'm currently doing this moving all here. I can take care of moving them in terms. I'm good too. Okay. So I just that way I'll keep them **Michael Baskin** [2:41](https://youtu.be/AvbsISUAIbU?t=161) vaguely. I know it's an extra. Yeah, I can. I don't know why. Okay. I can do this enough to be okay. So, okay. It's not. Okay. I think they're all out. They're all super. So, yeah. And I can I got it. That's it. Okay. Yeah. Perfect. Super. I think so. Yeah. **Unknown speaker** [3:15](https://youtu.be/AvbsISUAIbU?t=195) Andrew, welcome. As you join in, looks like there's an otter AI that needs to be moved out. **Michael Baskin** [3:22](https://youtu.be/AvbsISUAIbU?t=202) Oh, where I can see it during I do again. There you go. **Unknown speaker** [3:31](https://youtu.be/AvbsISUAIbU?t=211) Okay, all right, suis超. **Michael Baskin** [3:38](https://youtu.be/AvbsISUAIbU?t=218) Okay, move to and the other opportunity is when when I'm pulling out of breakout rooms, it's a good opportunity. I will take out the I will need to play. I need to explain that to people. Okay. That's okay. Great. Oh, you can we can do that. Okay. Let me check in the participants. So I will kick it off. **Unknown speaker** [3:55](https://youtu.be/AvbsISUAIbU?t=235) If you need any help to I'm like, I'm a fly on the wall. But if you want to just add me as co host if I can help with anything, let me know. Okay, just I'm doing you cause right now. **Michael Baskin** [4:05](https://youtu.be/AvbsISUAIbU?t=245) Wait a sec. I we are kicking off the note takers, because I'm not comfortable that I, I'm going to bring everyone back. Make a host. Okay, Jess, you're a good host now. Thanks, Jess. Awesome. Thanks. Wait a sec. I we are kicking off the note takers, because I'm not comfortable that I, I'm going to bring everyone back. Make a host. Okay, Jess, you're a good host now. Thanks, Jess. Awesome. Thanks. Great. Well, welcome everybody. If you want to drop into the chats, some of your experiences so places where you've practiced that will help build out our session for later. And times when you weren't ready so things that you shared yourself, or that you've heard from others. [5:06](https://youtu.be/AvbsISUAIbU?t=306) So welcome to Jenny I as a practice partner, there will be lots of breaking out in spaces here for those who have played in workshops with me before. So we're here to introduce sort of a new idea new concept, and to begin to build that concept out together, and then welcome all of you to contribute to building that going forward. So, thanks to innovate us for hosting this. And really glad to be able to build with a like minded community of public servants who are figuring out new ways to make stuff better. So I'm Michael I work in Montgomery County government towards more just human and liberating governments, and we're really working here to make sure that our government is more agile and resilient and adaptive. Some of our programs are innovation accelerators our flagship we've been launching it for six months before coded. And it meant that when our 9000 employees all went to Microsoft Teams 50% of our teams users were folks who had gone and practice this new thing which was Microsoft Teams. And that was in contrast with the rest of the organization that was waiting for a really big former rollout. [6:08](https://youtu.be/AvbsISUAIbU?t=368) And the theme we're going to sort of keep playing with is how do we have our organizations, jump in and practice more so they learn and are more ready for different features that emerge. I'm also a coach, and I'm a collector of outdoor public benches, which I just like to ask like fun questions about, like the one on the far right. If it's that three chairs or is that a bench. If you want to wander around cities with me and non virtual spaces that just a little bit about how I like to play. So today, our goal is to really understand the role of Jenny as a practice partner distinguished from other uses of generative AI to explore what makes this possible and then some what futures might be as a result of, of, of this new sort of approach. I know that we've got a lot of emergency preparedness folks here. First responders. This is not about emergency planning, it's about getting prepared. And I'd love if you're an emergency response professional if you just identify yourself in the chat or raise your hand and wave so we can see. [7:12](https://youtu.be/AvbsISUAIbU?t=432) Cool. Awesome. So we've got some people who are like pros at this, and then the rest of us who you have to deal with when the emergencies happen. So this is about the practice spaces for all of us together to be more prepared, right like we don't just look at game tape all day, or look at a bunch of plays, it's going out and practicing and getting on the scrimmage field that we actually learn and get better. So, we're going to be playing with Jenny as a practice partner you're going to be breaking back out into those groups. And then we're going to sort of work together to figure out what's coming next for this space. So we've already done some connection space, we are going to get talk about practice as a way to get be be more prepared to be more ready. Introduce Jenny as a practice partner, we're going to jump in and practice, and then we'll reflect together. So, practice. I just want to take everyone, take a moment to think about practice in your own life. And I welcome you all to share that into the chat some experiences when you've practiced where it's been useful to you. [8:17](https://youtu.be/AvbsISUAIbU?t=497) Maybe you were on sports teams growing up. Maybe you've had a mock interview that you wanted to get ready before you got your job in government. And for me I think a lot about bicycle riding right like you can sit and you can talk about riding a bicycle as much as you want. But the way that you learn how to ride a bicycle is by getting on that bicycle and actually writing and practice is super powerful because it builds those muscles and it builds that muscle memory. That's individual muscle that's organizational muscle right it allows us to figure out how does this actually work in reality I can see the mechanics I can read about the bike, but I'm going to jump out and give it a shot I can see this plan. Now I'm actually going to test this plan out. I'm going to figure out what's missing what's too much. And as I sort of do that work. It's creating these new neural pathways, it's creating muscle memory. It's building these habits of response so that when new things arise. We're ready to respond and not just to react. Even when that situation plays out differently than what we expected. If we're practiced, we're a lot more ready for that unknown. [9:22](https://youtu.be/AvbsISUAIbU?t=562) So it's all about sort of readiness of the individual and the organizational level. And I think practice really works when we take that action. And that reflection we combine it to create experience. And we have to realize that we don't have to like wait for something to happen to have that experience we can practice to begin to get more ready and more prepared. And I think really critically, that the, you know, practice spaces are learning spaces, which are spaces that we don't have as much of you know service as Tim mentioned it's the opportunity to fail to iterate to develop a system. So we know that we're practicing people act in slightly different ways right encourages people to make mistakes at an earlier stage, it allows us to try out new approaches because we know we're going to run a new in different practice session. Yeah, it allows us to do as john notes some co design work. And we can sort of test those plans to test those theories of change and those assumptions and practice a little bit different than drilling right drilling is sort of a component of it that's sort of like let's wax on let's wax off the karate kid thing. [10:28](https://youtu.be/AvbsISUAIbU?t=628) But it's really here moving from the drilling let's the drilling to that sparring. Let's play in live spaces with these scenarios, let's get on the field with a with another scrimmage team. So for me this became really important. I was in the mayor's office in Chattanooga on July 16. When we had a domestic terror attack in our city. It's a day that many of us will will never forget. And the police were ready, although they were screaming back because they are halfway to Nashville the chief of police was back with halfway to Nashville on the time. Right, the people who are in communications who are in the mayor's office with a terrorist in this case was a family member we learned quickly had a family member who worked in the department. How is that going to work with that department, what were we going to do. And the next year we had a school bus overturned another really tragic accident. We're more prepared, but our practice had been the terrorist attacks, and it really got me thinking like how do we make sure that we're more ready, not just that we have these plans because in the middle of the action it's not the best time to be dusting that plan off the shelf and testing it through for the first time. [11:40](https://youtu.be/AvbsISUAIbU?t=700) And we know that living through a thing isn't the only way to boost that level of experience. In government, you know, an outside of government we face these all the time since releasing the article, you know, I was thinking at the beginning about the the building collapse in Miami, but in wildfires in LA. But there's been so many things that we couldn't have imagined emerging just since just since the initial release of the practice as a partner article. So what I challenge folks to think about for practice as well is what are those greenfield opportunities, the gifts that come our way and how might we be more ready to catch them. And so I'd welcome just because it's going to warm us up folks to drop into the chat, some other opportunities for which folks were often not prepared. So think about what are the spaces we might not be prepared for. The key thing I just want to stress is that being planned like having a plan is not the same as being prepared. Justin, yeah, a parent, Justin and I were having a conversation yesterday about becoming parents. [12:43](https://youtu.be/AvbsISUAIbU?t=763) There's no plan for that right but there's things that you can do to practice strengthening your relationship with your co parent to get ready for it. Right. That's a great example. Thanks, Justin. You know, there's a great quote that says, everyone has a plan until you get punched in the face. That's a Mike Tyson quote. So when we think about practice I think the thing that we want to be asking is how do we get punched in the face as soon as possible. And the way to get punched in the face it's kind of hard to punch yourself is to get yourself a scrimmage partner, a sparring partner who can jump into the ring with you. Yeah, Giuliana and illness. Yep, those are great ones. Often I think one of the challenges with planning is planning can lead to complacency it's it's practice that really leads to competence. And so in the past, the main way that we've been doing this is tabletop activities tabletops not just what you need an exercise or a yoga mat for in 2018 we brought together a bunch of city census leads because we want to get them ready for things that were going to be happening in the census. So we had to pull everyone together with some amazing support from the data and society folks they're real experts in the space, but required experts in the subject matter, as well as experts in pedagogy and how to actually get adults to learn in order to pull this together to create these tabletop activities and exercises. So this practice in the past has been really costly and as a result, we don't do it. [14:08](https://youtu.be/AvbsISUAIbU?t=848) It takes a lot set these activities up. And they're often just not realistic enough. Before Jenny I get to imagine all of those sets of possible paths and there was always ones that you hadn't actually imagined, and it was really hard to update. Once you've gotten started. Yeah, we're totally unprepared anybody nice to see you unprepared for some shifting demographics. And what in those great those are some bigger scale things we might want to practice for. But Jenny I have some really unique properties that create a new way forward. So, you know, here's what this sort of looks like in practice, I'm going to show you two models. One is it's just Gemini of course at innovate us it's always. You can use prompts I've provided, but this is always useful just to think about. Any tool. So we're tool agnostic here. So we just push in a prompt. As usual, it's helpful to be specific when directing it. [15:11](https://youtu.be/AvbsISUAIbU?t=911) So you're going to run a scenario tabletop activity to help us better prepare for a bus crash incident. First, generate one scenario that we can provide to the team. Then we'll have you run through a live activity with them providing updates as we respond. You'll prompt us one by one in a set of questions that we have to respond to. You can help us respond, but get us to do the work. Throw in some updates to the scenario that respond to the solutions and plans as we go. So just always good prompting advice. Being specific, providing some context, giving ourselves a role and giving it a role. So, you know, we're a small city manager's team, including city manager, comms director, chief of staff, and the role that it's going to play. Maybe providing examples. So if you're an emergency preparedness professional, you might drop in your EOC plan. Or if you already have tabletop exercises, you might give it an example. I've also created a custom GPT, which we've just rolling out today. I'll drop it into the link for our practice sessions that you can use to play that is primed up to create these scenarios and to run you through these scenarios as a practice partner. [16:15](https://youtu.be/AvbsISUAIbU?t=975) So this is what it looks like. It'll share, you know, here's the scenario that begins. It's going to look different any time. That's the fun of generative AI. And I just want you all to just take a moment to actually do this, actually step into this. So read the scenario and type your responses into the chat. And then I'll hit, I'll say, go. I'll hit enter. So it's 830. It's a Monday morning. Your city's IT team has just informed you that ransomware has locked critical systems. I'm sorry for the folks in the room from Atlanta and Baltimore for whom this is all too real. Attackers remaining 5 million in Bitcoin to restore access and only Miami has Bitcoin on hand. Meanwhile, misinformation is already spreading online. Claims their administration ignored warnings about cybersecurity vulnerabilities. You got one hour. Here's our first prompt. So let's go for it. I want you to feel that experience. So go ahead, type your initial response into the chat. [17:23](https://youtu.be/AvbsISUAIbU?t=1043) And so the fun thing here is you need to actually respond to all three of these questions. Good. Start with three priorities. So now you know, actually, Erwin, this is great. I want you to figure out what are the top three priorities that you need to address in the first hour. Great, Juliana. Nice. Good one. Nice, Norbert. Awesome. Thanks, Allison. Nice, Bab. You've been in the mayor's office a while. Cool. And what I hope that you sort of think about in this space is what was it like when normally we just get to read or look at plans when you're actually asked to respond live and in that moment. And so one really cool way that we can use Gen AI as a practice partner is switching our tools into live mode. So we were doing, we were imagining that we'd have federal agents reading our buses in Montgomery County. [18:27](https://youtu.be/AvbsISUAIbU?t=1107) And so we were going through this scenario and we realized, all right, Governor Moore is calling. He wants an update because there's crowds gathering. What do you do? Governor Moore is now on the line. We switched into voice mode. AI took on the role of Governor Moore. And the first response, I said, OK, pick up the phone. You're there. That a member of our team had was, oh, wait, wait, wait. I'm not ready. And that was actually a super powerful real response because that's often how we feel when we're put into that situation of actually getting onto the practice field. It's a totally different mental shift from just being able to sit and write out a plan. Cool. So just take a moment to reflect on what it felt like to be put on that spot. That's that first punch. And the more it happens, the more ready that you are. And then the fun thing about Gen AI is this is a live model. It then gives your next development and then your next development and then your next development. And so when we use generative AI as a practice partner, what we're doing is we're allowing to set up those scenarios, going through some rapid iterations of the team responding, the scenario response. [19:31](https://youtu.be/AvbsISUAIbU?t=1171) And that's the powerful part about using generative AI, right, is that these are open, complex models where the team responds, then we respond, the team responds, then we respond. It can then help to identify, to evaluate that response, to provide some feedback on that response. Here are some different alternatives. You might feed it based on it. You might feed in your current EOC plan. And it might then say, hey, you forgot about bullet points 7, 8, and 9. You might want to strengthen that up here. Or we suddenly had more disinformation than we were prepared for. Or based on your response, here's how you might want to rearrange your plan. And maybe you can build some smart, quick sheets. That's using it not as a practice partner, but as more of a compliment. And that can help us prompt reflection. So it can begin to ask those questions of your team or your group or you, whoever's using that practice partner, to reflect on their experience. And then from there, you set up a whole new scenario that it's then setting up for you and running you through again and again and again. And I think what's really sort of key or critical here is in the old model of doing tabletop exercises, scenario-based plans, full-on simulations, the best that we really had was we were taking these really complex scenarios and we were just reducing them down to complicated games. [20:47](https://youtu.be/AvbsISUAIbU?t=1247) But the world we know is complex. And so what we were doing is trying to map out all of these scenarios in advance. But you were choosing. It was instinctive for some of you in the chat to say, oh, just choose option B. Choose option D. And we're really good at those choose a multiple choice answer exams. But the world isn't a multiple choice answer. And generative AI as a practice partner can get us beyond that. In some ways, the old tabletop exercises to me always sort of felt like those goosebumps stories, those choose your own adventure stories. And they were really expensive. The mistakes were costly. The cost of a do-over was costly. And so what we would often do is sort of engage our analysis brain instead of our action brain. And I think with generative AI, we're allowed to create some more complex scenarios that are responsive to us. And a way to think about this, and I borrow this from Fred Lelew of Reinventing Organizations, is like our old models were really, really good for engineers for building like a whole Boeing airplane that you could take apart piece by piece and have it constructed and ready. [21:55](https://youtu.be/AvbsISUAIbU?t=1315) But life is more like a bowl of spaghetti where when you pull on a single strand, you don't really know what's going to be happening next. There's so much complexity in that space. And with generative AI, these tabletop activities are not just more robust, but they're cheaper. They're easier. You can run them multiple times, so mistakes aren't as costly. You can do a series of many tests that build up to a big simulation instead of one big simulation. Here are some of the things that have emerged. I really appreciate the feedback that people have been using this as a practice partner, this from-to space. And I'd love, as you all play with this, to be thinking about this and welcome folks to drop into the chat. But the old models were really, sometimes it was scary because you were jumping into this space versus feeling really accessible. You can practice on your own first. They were static. They were pre-built versus being really flexible and responsive. They were really costly. And now with generative AI as a practice partner, they're much cheaper. [22:58](https://youtu.be/AvbsISUAIbU?t=1378) They're much more inexpensive. It takes much less time to build them and less expertise to build them, but allows expertise, especially from emergency folks, to be leveraged in stronger ways. In the past, you tried to do one well. With generative AI as a practice partner, there's a lot of room for action, play space, room for creativity. And it really encouraged us to step into that rapid prototyping. Adults don't really love to fail. That's something we've been trained into. This allows us to step into that failing space. We tend to have lots more of the old tabletop activities. We'd have lots of meetings about that. And we'd have lots of talking about that. And this allows us to go from talk and consider to decide and do. And that's what helps build our competence and makes us more ready. So I just want to fit this into the frame of how we use generative AI before we jump into some practice. And a lot of ways that we've currently been using generative AI are in a replacement model. So it's going to do things for me. [23:59](https://youtu.be/AvbsISUAIbU?t=1439) That could be it's writing email responses for me. It's going to generate a full policy brief with limited input. It might even be drafting scenario scripts. It's going to do some replacement, do it for me. And some folks have been thinking about it also is going to help me do stuff. So it could analyze those scenarios for us. It could roll out the scenarios and do all the work for us and help me think through all the things that could happen and present those pros and cons. Lots of us are using it for first drafts or for editing. It can help us prepare agendas for these types of things. There's some folks I think Ethan sort of talks about using as a conversation partner and help me think. And I want to propose that using as a practice partner is something that's sort of distinct and different. Because rather than it doing it for me or helping me to do stuff, this is less about the generative AI and more about the human intelligence, those human muscles that are getting stronger. So it's how do we use this as a tool not to get things done, but to strengthen our ability to get things done now and in the future. [25:05](https://youtu.be/AvbsISUAIbU?t=1505) So when we talk about using generative AI as a practice partner, we're talking about it where it's engaging me in an evolving scenario or game. It's providing that dynamic feedback back and forth. It's helping to run simulations that are responsive to my actions. Lots of folks now are using it for mock interviews. Or if you're preparing for a presentation in front of a board or city council and you want to get grilled, it's going to grill you in advance. Using it as a counterparty in negotiations, these are spaces where you're sort of going back and forth with it. It's not just doing the work for you or helping you do the work, but it's literally helping us to get stronger in doing the work. That's sort of the idea. I think it's worth experiencing it, practicing it, and giving it a try. So two things here. I'm going to drop into the chat a link to a custom GPT that is set up to guide you in setting up a practice scenario. You're also welcome to use any tool that you want. And so what you're going to do here is you're going to jump into these groups. [26:10](https://youtu.be/AvbsISUAIbU?t=1570) Yeah, that's a great idea, Juan. So Juan is using this with our Office of Food Systems Resilience to help imagine. He works with lots and lots of partners. He's going to react in this space. You can do it, Maggie, yes, without logging in. And you can use any tool that you work or like. It's also totally fine to just have just one person in the breakout room using the tool. So in small groups, you are going to be choosing a tool, set the scene. If you're using the custom GPT, it'll sort of prompt you to help set up a strong team. I'd also encourage folks to just jump into Gemini or Cloud or whatever tool you like and use one of these practice prompts. You're going to go ahead and play in practice. We're going to have 15 full minutes to go and do it. OK, Maggie. And then iterate and do some reflection. Just so there's no Zoom whiplash for folks who haven't done one, two, four all before, let me just walk you through what happens. At some point in the middle of this great, fun playing space, which is great, so about 15 minutes in, I will broadcast a message. [27:17](https://youtu.be/AvbsISUAIbU?t=1637) You're then going to find yourself just, you can go off screen for a second, one minute of silent reflection for oneself. You're asking the question, what happened? What did you learn? What's next? What happened? What did you learn? What's next? And then we're going to use some funky breakout room space. So from single person reflection, you're going to go into twos. Those twos are going to get knocked together into fours to see themes across. And then we'll come back together as an all to reflect as a group. Any questions, comments, concerns, accusations, or limitations before we throw you out? I will jump from breakout room to breakout room. And you can always jump back into the main room if you need or want. And I'm going to try to keep the rooms as they were before. But we had some so that you know the folks in there. And I'll do some adjusting. If you're in a room by yourself, hang tight. [28:18](https://youtu.be/AvbsISUAIbU?t=1698) Because we kicked out a bunch of AI folks. And you'll be back together. Suzanne, you are in a room by yourself, ironically. And so you're going to go to room seven. All right, off we go. And we'll be back together shortly. Let's give it a shot. **Audience** [28:36](https://youtu.be/AvbsISUAIbU?t=1716) Michael, loving absolutely the workshop. But I mean, crazy loving the workshop. We need audio too. **Michael Baskin** [28:54](https://youtu.be/AvbsISUAIbU?t=1734) OK. OK, there we go. All right, and I'm going to do the room. So if anyone else is messing with them, give them a second so I can knock it out. Oh, we got a bunch of people who are unassigned. Hey, awesome. OK, so a bunch of people joined late. So I'm going to give me half a second. I'm going to add you all to rooms. [29:54](https://youtu.be/AvbsISUAIbU?t=1794) That'll just take a second. And welcome. If you give me a half a second, let me do it, otherwise it's going to mess up Zoom. That's okay, Samantha, I'm going to join you into a room anyway. And whoever called in with a 9518 number, if you let me know what your name is, I can add you to the room where the rest of your person is. [31:11](https://youtu.be/AvbsISUAIbU?t=1871) Great. **Audience** [31:23](https://youtu.be/AvbsISUAIbU?t=1883) I was just going to add into the chat, I just popped back from one. I think at least my group is a little bit stuck and wondering if through the announcement feature you can send some of the prompts and some nudges, just since we landed in a totally empty space and are all arriving and orienting, so any of the visual slides or prompts would be helpful. Now, can I get back to my breakout room? If not, let me at one open up, and then two. **Michael Baskin** [32:09](https://youtu.be/AvbsISUAIbU?t=1929) Thanks, Sophie. Yeah. **Audience** [32:15](https://youtu.be/AvbsISUAIbU?t=1935) Sorry, I was just in group 12. Yep. If you're able to send me back. **Michael Baskin** [32:22](https://youtu.be/AvbsISUAIbU?t=1942) Patrick, Patrick, you got it. Thanks, Patrick. **Audience** [32:26](https://youtu.be/AvbsISUAIbU?t=1946) Thank you. Yeah. Okay. **Michael Baskin** [33:43](https://youtu.be/AvbsISUAIbU?t=2023) I'm going to jump from different rooms. [36:14](https://youtu.be/AvbsISUAIbU?t=2174) Have a good one. Bye. Bye. Bye. Thanks, Patrick. Bye. Bye. Bye. [37:14](https://youtu.be/AvbsISUAIbU?t=2234) Bye. Bye. Bye. Bye. Bye. [38:14](https://youtu.be/AvbsISUAIbU?t=2294) Bye. Bye. Bye. [39:14](https://youtu.be/AvbsISUAIbU?t=2354) Bye. Bye. Bye. Bye. [40:14](https://youtu.be/AvbsISUAIbU?t=2414) Bye. Bye. Bye. Bye. Bye. [41:14](https://youtu.be/AvbsISUAIbU?t=2474) Bye. Bye. Bye. Bye. [42:14](https://youtu.be/AvbsISUAIbU?t=2534) Bye. Bye. Bye. Bye. Bye. [43:14](https://youtu.be/AvbsISUAIbU?t=2594) Bye. Bye. Bye. Bye. [44:14](https://youtu.be/AvbsISUAIbU?t=2654) Bye. Bye. Bye. Bye. [45:30](https://youtu.be/AvbsISUAIbU?t=2730) Awesome. Cool. Awesome. Here comes the Zoom whiplash. [46:43](https://youtu.be/AvbsISUAIbU?t=2803) There you go. **Audience** [46:44](https://youtu.be/AvbsISUAIbU?t=2804) Hey, Michael, are you willing to like share how you made that custom GPT someone was asking about that? **Michael Baskin** [46:51](https://youtu.be/AvbsISUAIbU?t=2811) Yeah, I can share the full back end. I'm happy to do that. **Audience** [46:57](https://youtu.be/AvbsISUAIbU?t=2817) Yeah, that would be super helpful. Thank you. **Michael Baskin** [47:00](https://youtu.be/AvbsISUAIbU?t=2820) Okay. Cool. I'm gonna give folks. Let's see if this works. We've got enough folks, I'm just gonna close all the rooms here. And we're going to come into one big space. Awesome. Cool. Just any quick, I'm gonna, it's I have a question of how we sort of build this on the back end. Any quick reflections on what worked, what didn't work for folks in using Jenna as a practice partner? You want chat GPT plus. So invest the, invest the money. Yep. [48:01](https://youtu.be/AvbsISUAIbU?t=2881) Using a plus tool is helpful. Yep. **Audience** [48:03](https://youtu.be/AvbsISUAIbU?t=2883) Oh, go ahead, Michael, I'm sorry. **Michael Baskin** [48:08](https://youtu.be/AvbsISUAIbU?t=2888) Go for it, Betty. **Audience** [48:11](https://youtu.be/AvbsISUAIbU?t=2891) Yeah, in our team, we were a group of five. And I didn't see on the chat that you had a practice. You put your chat GPT, so I do pay for it. And it's the magic is on the prompts. So we were looking for the first two questions. But the hardest part was deciding on the scenario. You know, what was it that we wanted to be specific about for it to help us? **Michael Baskin** [48:37](https://youtu.be/AvbsISUAIbU?t=2917) Yeah. Yeah, I don't know if anybody from our team wants to share. I think it's awesome. I think it speaks in some ways to what adults want to do, which is we really like to talk about the issues. And what we try to avoid as much as possible as being thrown into a space where we actually have to respond, we actually have to practice. I didn't see any groups do what I often do, which is, I'll simply feed the prompt from one into another, and then it's playing the scenario out for me. But then I'm using as a compliment to me, right? It's doing some of the planning and work on my behalf, instead of using as a practice partner, which is where it's actually forcing us as a group. Once we get in our computers, we have to get stuck in our computers versus here's the prompt. Here's the tabletop. Let's figure out as a group what we're going to do. Let's put a response in. Let's see what happens next. Did anyone use a live audio mode? **Audience** [49:33](https://youtu.be/AvbsISUAIbU?t=2973) Oh, no. That would be so cool. **Michael Baskin** [49:37](https://youtu.be/AvbsISUAIbU?t=2977) Yeah. It's a really sort of good, scary way. Thanks, Vicki. Kristen, what would you like to hear more about? Awesome. So as a practice partner, we're really thinking about, yeah, is, right, or when we're using it to provide suggestions for what we could do, then we're using it really as a compliment, right? It's going to help me do this thing. And that's an awesome way, right? Like when we're in real life scenarios now, if there is a real building collapse that's happening in Montgomery County, I probably do want to be using chat GPT or some other tool to generate a bunch of the responses that are going out for me. It's going to speed things up. That's using as a compliment. How we want to be introducing this is using this as a practice partner, which is then us thinking, all right, what are we going to do? Who needs to be in the room? Oh, we need to put out some extra tweets. Let's go generate them, et cetera. [50:37](https://youtu.be/AvbsISUAIbU?t=3037) So we're often blending the different ways that we're using these tools. The distinct way to think about this is, yeah, really as role play, thinking about it as a scrimmage partner so that you're actively acting out the scenario and going through that tough space. Yeah. How to feed one into the other. So I can show you that. **Audience** [50:55](https://youtu.be/AvbsISUAIbU?t=3055) I think we started doing some of that in our group in breakout room nine. And I think one observation is just there is in general like a high barrier to entry, whether that's due to just like familiarity or even security issues with your agency and whether you have access to things. And I think that was just an observation that made like getting started kind of difficult. And so I just think as we want to scale adoption to like de-risk our own behavior, there are also just basic like infrastructural blockers that make this kind of difficult. And I think we experienced some of that in our room. I think once we put them in different models, we could see, okay, there are differences between chat GPT and cloud that Megan captured in the chat. I think ultimately we were going to do the next step, which was still call 911, even though we had different levels of detail provided, very drastically different levels of scenario, quality of the scenarios provided our immediate next step was the same. [52:06](https://youtu.be/AvbsISUAIbU?t=3126) But I'd be curious to like let the scenario play out longer and feel free to jump in group if I simplified any of that or missed anything. **Michael Baskin** [52:13](https://youtu.be/AvbsISUAIbU?t=3133) I love it, Sophie. This is definitely sort of, it takes, you have to be comfortable already using these gen AI tools to make the most of this as a practice partner. Just a new way of sort of thinking about it. So I'm happy to share, this is the backend. I can't copy and paste the full thing, but I can drop it into the slides, which are shared out later. But this is the backend of the custom GPT so that anyone can recreate it and to sort of see what that looks like. So what you could do is you can just pull in the response from one and drop it into the other. So. So that you can then have it respond. [53:18](https://youtu.be/AvbsISUAIbU?t=3198) And that's, this is one way to, and then copy it right over into the practice space and go. The key thing here is we're using it. So this is a cool way to sort of play out a scenario. What you're missing in doing this is you're letting gen AI do the practice for you rather than doing that practice yourself. And so the question is, are we strengthening our muscles of responsiveness or are we weakening them? But this allows you to see if you were running the scenario for a group or a team, sometimes it's helpful to let the whole scenario run out. Great. And then we'd simply copy it and paste it back in. And you can let those two folks together. So just showing how that works. Awesome. We're close on time here. I want to thank everyone for joining in. I think sort of the big why here is just how important it is to jump into these practices, this habits of action and learning and jumping into iterative cycles. [54:25](https://youtu.be/AvbsISUAIbU?t=3265) If we want to have more adaptive and more resilient organizations ready for these emergent futures that are less brittle and more ready to serve people where they are, the question is how do we say, and we just have an hour in this meeting, how do we get this whole team to run through practice scenarios? As I think some of you found having one person running the practice session and the rest responding can be really helpful. The more government fails and the more that we can fail faster and learn more, the more that we are in a good space as public servants to step into our agency and step into our power and build better work for those by we serve. If folks want to drop their emails or LinkedIn's into the chat to continue the conversation or just sort of release this live and so later folks come on, join in. If you want to drop your comments, they will die in this chat, but if you drop them in response to the link I just shared, then that will continue onwards. Awesome. Thanks so much. And I really like we're still developing this practice partner concept and would love to hear how people are using it, what's working and how we can continue to make it better to strengthen governments and make sure we're more ready for the futures to come. [55:30](https://youtu.be/AvbsISUAIbU?t=3330) Thanks so much. **Audience** [55:31](https://youtu.be/AvbsISUAIbU?t=3331) Thank you very much. And thank you, Michael, for this impressive workshop and all the participants and hope to see you soon in our next workshop. Thank you. Thank you. **Michael Baskin** [55:41](https://youtu.be/AvbsISUAIbU?t=3341) Thanks, Marco. * * * Original page: https://innovate-us.org/from-plans-to-preparedness-using-gen-ai-as-a-practice-partner-in-the-public-sector · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org