Prompting the Future: AI, Leadership, and the Military Mindset with Capt. M. Scott Austin, USCG

Yeah. Ford, Carolyn 0:56 But you get it wrong. Or you're vague, like remove all obstacles and now you're missing walls and doors, and maybe senior leadership and so. It's. Incredibly important to be precise and clear, and today we'll dive into why getting the prompt right isn't just important, it's mission critical. So welcome to the show, Scott. Thank you. Thank you so much. Ford, Carolyn 1:22 Well, let's in your article. So at first I found you from your article in Signal magazine, which I'll drop in the show notes and in the article you highlight the importance of framing the right questions to effectively utilize AI. How can military leaders develop the skill of prompt engineering to ensure that AI tools address specific operational challenges?

Hey, thanks. I guess the first thing to clarify, make sure in aviation. A tech aficionado, but my most of my career has actually been in the aviation community. So in in aviation, we say you should always communicate and be accurate, bold and concise. The ABC's of communicating right. So I'll try not to be too concise and make for pretty boring podcast, but to be accurate, the term front engineering in the article was kind of used in a bit of word play. The idea of saying the article isn't going to tell a senior leader that reads it how to write an effective AI prompt. The article was born from witnessing some challenges within the capability development side of the military, particularly on the DoD side. But it's it's for me. It's pretty much across the board where AI is very popular. Leaders know that AI has a lot of things that it brings to the table, and so they say go get me some of that AI. Whatever the problem is, I want some AI. You know that. You know the vanilla ice version. You know, if you got a problem, AI is gonna gonna be the solution and then we'll find the problem. And we don't that in other fields.

We start with the problem, but when it comes to AI, there's this, this novelty, this lack of understanding and a lot of senior leaders. I think we're jumping to the AI conclusion. So the prompt engineering idea is to say, hey, it's a little bit of a design thinking idea, right? The idea is. Start with thinking about the user. Start with defining the problem. That you're trying to solve. And then go from there to identify what solutions might solve it. And AI is going to be one of those problems that fits in a giant toolbox of things that you consider. But it's only one of those things. And so prompt engineering the things that make a good prompt engineer right. The principles like you said on the Starship Enterprise. Clear direction. Providing some context iterating on an idea towards getting a final answer, but always starting with the problem. That you're trying to solve is kind of the idea of how you use prompt engineering to make sure the AI is rolled out in the most useful and not overused way, I think. I mean I I asked before this. I asked. I hopped on to a ChatGPT and asked hey, what makes a good prompt engineer, right? And it it speed out a few things. It said things like.

Communication skills, empathy, and understanding. Structured thinking. And curiosity? And then only the last thing that it mentioned was some sort of technical literacy or a technical aspect. And it's interesting that that's lowest on the rung. The key there is you know that empathy with your user, that ability to identify what the real problem is and dig into some problem solving skills. Those are the things, hopefully one that the leaders have already used to get into the position that they're at. So they don't need to be scared of this AI tool. They can use those same skills that they use in the other places. Just don't forget to bring those to the table when you're talking about AI.

Ford, Carolyn 4:57 Yeah. Well, and you know, the word engineering is a little bit of a misdirect because it makes me immediately think, well, you got to have these technical skills, but. You know, as I'm using AI to help me write a better white paper or or whatever it is that I'm writing in marketing, I find the more to your point that I. The more context, the more detail. A lot of times. With AI, the beauty of it I can just do like word of vomit. And like, work out what I'm trying to get at, but I share it with the AI and then I ask the AI to, you know, do you understand me?

Mm hmm. Ford, Carolyn 5:43 And then iterate and go back and forth and actually collaborate with the AI. But. I. So in the article. I think you're talking about not only actual literally AI prompting, but. Being able to talk about AI and educate about AI, about what it is and what it's not. So you're telling me it's not magic and it can't solve all the problems? Right. Oh, shame all some of that hype isn't isn't as found as founded as we had hoped.

Ford, Carolyn 6:22 Yeah. So, drawing from your diverse roles in the Coast Guard, how can leaders balance the adoption of cutting edge AI technologies with the realities of organizational culture and existing workflows? I see the bottom line to me is provide value. Right. Back to the the first point a little bit is start. Empathy is the first step in using or deciding if you're gonna use these kind of tools, you have to go to the person that's doing the job. In that design thinking process that you talk, we talk about and it it starts with seeing what the problems are that. They're really dealing with on it. What would make their process better, right? Because you can add things to the process and if if they don't help, they. Wonder. Never going to be adopted.

Ford, Carolyn 7:19 Mm hmm. They don't help. They're not going to be adopted, but you have to meet the users where they are. I think you had a while back. You had a podcast guest that talked about using empathy monitoring with empathy, right? Bill James. Ford, Carolyn 7:36 Mm hmm. And it brought to mind something from our community, again, from the aviation world. The idea of personal investment. And part the tools that you're providing and then maintaining. So in the Coast Guard, our maintainers, the folks that work on our helicopters are unique. Unlike other services, the folks that work on the plane are the same ones that fly on the plane.

Ford, Carolyn 8:03 Oh. So there, yeah. So in in other fields, you know in other services there are large enough, a lot of the folks that are working on the helicopter are fixing the plane. They they stay primarily on the ground when the plane takes off. There's somebody else whose job is. Ford, Carolyn 8:15 I had no idea. Yeah. So we're we're, yeah. Ford, Carolyn 8:17 I had no idea that you guys were mechanics and pilots.

Oh no, not the pilots, but the the folks that work in the back of the plane. So the flight mechanics, the folks that the crew primarily are the folks. Ford, Carolyn 8:25 OK, the crew. So there's a personal investment. You know, the same plane that they're repairing the same plane that they're maintaining is the plane that they're going to be in doing their job when they're out operating, right. And so there's a. There's a personal investment in providing a tool to the job that's going to solve the job that's going to be useful. For the operation, I think y'all talked a little bit about. That's where a user interface has come to mind. Whenever you're talking about designing, right?

Ford, Carolyn 8:55 Mm hmm. An engineer feels great. Hey, this this thing I've I've solved the problem. I've written the code. It goes on the screen. It's your problem. If you can't interpret what those dots and lines mean, right? But if you start with the the the user, the person that's actually doing the operation or using the tool, you'll identify kind of low hanging fruit. That can provide value. Sometimes it's not a complex, you know AII puzzle that engineers enjoy solving. Sometimes it's something that's so kind of below their skill level that they kind of roll their eyes, but that provides the most value. We here at CMU. I was lucky enough in this program to talk to Alexis Bonnell. She's the CIO. She was the CIO for Air Force Research labs, and she had a great focus that she talked about about buying back time. Using artificial intelligence is a key tool to buy back time for the workforce.

Ford, Carolyn 9:52 Yeah, right. And she talked about the way to do that is to eliminate toil, which was one of her big phrases. She said. Toil kills purpose faster than the mission can replace it. So if you think about all those things in your day that you do the the checking e-mail, the curation of the e-mail, the things that suck exactly right. Ford, Carolyn 10:09 Sucks your soul. It sucks your soul, Scott.

So, so I mean, military leaders. They might say, hey, I need AII want it to go. I want to solve these massive, you know, robotics and you know. Huge projects when a huge part of our workforce is just screaming like, Oh my God. Just have ChatGPT on my military work station. Ford, Carolyn 10:31 Yeah. Great. Can it just do my expense report for me please? Could you do something about my e-mail? Exactly. Oh, if I could, if I. If I could get rid of travel claims.

Ford, Carolyn 10:37 For the love of God, don't make me do one more expense report. You know the change. Ford, Carolyn 10:42 Yes, right. I could make to the right and and and that's the key. The key to what we're getting so excited about that right is because it's providing value and that's the key to adoption and that's the key to getting AI permeating into the people's Normal work day. Ford, Carolyn 10:53 Mm hmm.

Yeah, I mean we've we've done it on the outside, right? You just said nobody stuck a gun deer head and said, hey, you have to start using ChatGPT. It provided value, right? And and adoption is it's just a really it's a byproduct of that. Ford, Carolyn 11:08 Right. Yeah, and it allows me to do the things more of the things that I love to do because it frees up my time from those rote tasks. Yes. Ford, Carolyn 11:21 Tell me Alexis the quote again, you said.

She says. Oh, I love it, she says toil. It kills purpose faster than mission can replace it. Ford, Carolyn 11:33 Yeah. Because people signed up for our service to do the mission, they didn't sign up to do travel claims and curate e-mail, right? Ford, Carolyn 11:39 Right. And and we didn't hire them to do that. Really. We hired them, you know, to to fight our wars, to do the things that we you know that we had on the recruiting poster. And and that's that's providing value, I think that's a that's a huge point.

Ford, Carolyn 11:54 Yeah, yeah, I love that. So I'm gonna shift gears a little. Well first. The course that do you you teach a course at Carnegie Mellon? Is that OK? I do not teach right now. Ford, Carolyn 12:13 Oh. I lost him. Hi. Hello. I'm so sorry I got kicked off. Ford, Carolyn 13:10 That's OK. You know what the beauty of not doing these live is? It's no stress. I'll just edit that piece out.

Go and we'll ever know. OK. Ford, Carolyn 13:18 No big deal. I'll just tell my AI. You know that spot where Scott disappeared? Cut that out. Exactly. There you go. See, that's where agents will be great. Ford, Carolyn 13:26 Yeah. OK. So you were telling me what you're doing at Carnegie Mellon right now? You're not teaching a course. No. So I am here as a part of the Carnegie Mellon Institute for Strategy and Technology, the CMS program. So there's a member of each of the Armed Services that's here, and we do basically joint research for a year with parts of the staff, people within our program and then outside.

Ford, Carolyn 13:47 OK. So if we've gotten the chance to work on everything from automated drones to. Learning some things about design and decision science. All kinds of stuff I didn't get to take. The performing arts classes. But that was probably a a better idea. A better use of my time. Ford, Carolyn 14:10 Well, so it sounds like you use AI very similar to the way I use AI then. You're using it to collaborate on your research. Even when you write things. Yeah. So we use it very similar.

Mm H. Ford, Carolyn 14:25 I mean, what you do with it might be a little more than what I do with it, but. Hardly. Ford, Carolyn 14:32 So OK, I'm not gonna lie. When I when I read the article. Mm hmm. Ford, Carolyn 14:39 My brain immediately went to OK. How do I do? AI prompting better, like how do I learn to do better? Mm hmm. Ford, Carolyn 14:47 AI prompting and your article now that I get to talk to you. Like you said. It's more about how to educate on AI for AI adoption. So you were using that AI engineering AI prompt engineering as? How to talk about AIA little tongue in cheek? Yeah.

Almost tongue in cheek. Ford, Carolyn 15:14 So. I. I want to ask you, you said something right before we got cut off about. Well, just everything that you're doing at Carnegie Mellon now, I'm trying to remember, OK. We'll edit this part out. I want to shift gears to something that you just posted on your LinkedIn page. OK. Ford, Carolyn 15:38 It's an article that I read and then I saw you post it and it's the title of the article. Is agentic warfare is here? Will America be the first mover? You said. What do folks think is the nuance between generative, which is created creating an agentic which is doing? AI this year's overhyped buzz word? Or the next? Or the next big step. So I want to throw your question back at you. What do you think?

Of course. Well, I think. Ford, Carolyn 16:12 Is it a buzzword? It absolutely is, I think. But it's not just a buzzword, right? AI. What a year ago. Two years ago was and in a lot of ways it is. And that was kind of the point of my article in some ways to tie the two together is a lot of senior leaders view AI just as this buzzword, and they're easily dubbed. There's companies that love to sell them, things they don't need, and they love to come.

Ford, Carolyn 16:32 Mm hmm. To their people and and mandate. To their people use these tools. Use this AI that I've heard so much about that I know so little about. To solve your problem and you know the people are sitting there wringing their hands, saying thank you. But that's not what I need right now. Ford, Carolyn 16:49 Mm H. And so to take it to agentic AI, yes, it's it's, it is, I think the next big step, right. So the the novelty of being able to create. New emojis. Is gonna wear off pretty shortly, and I think what again, this is this is where companies are great people. Companies are doing what we're talking about and design things they didn't come to people and say, hey, we figured out how to make. A tool that can take your your word prompt and create something for you. They know that's not what you want in the end. That's not. That's solving one of your problems, but we've talked. Our bigger problem is I don't want it just to create something for me. I want it to do something for me. Right, so I want. I want the tool. I want the intelligence to say hey, here's the problem I presented to you and it can. It can't just fire that prompt back to so early.

A lot of early large language models, right? So you say, hey. Create this memo for me or or create something for me and the prompt the results that you got back, you would say hey, create this for me and it would give you back instructions about how to create it. Yeah, I don't know if you've ever had that experience and then, you know, if some of us are snarky and we fire back, you're like, no, you work for me. You create it, you know?

Ford, Carolyn 18:12 Mm hmm. And I was like, sorry, I don't know how yet. But now we're getting to the step for agentic AI can finally stop firing those things back to us, and that will be the next step in the process. And that's the problem that we all brought to the, you know, to this issue is, hey, I don't want a tool that can just. Give me words back. I want a tool that can actually again buy me back more time by taking these rote tasks. These sometimes very important tasks. In the defense world and and do those for me because it knows what the next step or steps are and it doesn't need me to do it for them.

Ford, Carolyn 18:58 So when we talk about the agentic the doing like I think about my own little R2D2 that just knows like, oh, that's broken. I'm gonna go fix it. And he knows how to do it. Like, that's awesome to me. Team Human Machine teaming, right? Mm hmm. Ford, Carolyn 19:21 As we think about it on the battlefield, and let's literally think about AI prompting in a very little literal way. Are we going to be prompting those agents or are they really going to be more R2D2, I guess. Luke and R2, he definitely prompts him, right? But it seems like sometimes R2D2's smarter than Luke.

Yep. Oh, it's fun to live in these analogies, right? Like so. I mean, you go to it, right? C3. Po and R2, they're they're a great lesson for you in in the so you know C3PO, he was a what are they called? A protocol Droid, right. So C3PO always does. Ford, Carolyn 19:59 So annoying. Literal yes. It's so literal, right? So early on, no snark in the model, right? Ford, Carolyn 20:04 Yeah. Mm hmm. R2 is famous for always doing what he thinks Luke needs, not what Luke tells him, right? And that always adds to the adventure, right?

Ford, Carolyn 20:17 Yeah. You know, happens in every movie and we never get tired of it. And I think and that's why everyone ends up loving R2, right? Ford, Carolyn 20:25 Right. So the question is, you know. What level of autonomy are you willing to accept? Ford, Carolyn 20:32 Are we gonna? Yeah. Right 'cause, this is the question, not necessarily about intelligence. It's about autonomy, and that's and that's in the tool that you define, right? That's all about. That's risk, right? I think there was AI mean we may talk about risk later, but. That's that's that's up to the decision maker and that's part of that process is determining is the is the utility is the time I buy back is the time worth the risk. And in some of these cases, and we're we're finding already in artificial intelligence systems, hey, the because we're using one system against another system, we have to accept that risk because the other system operates faster than we can with the human in the loop, right, you talk about.

Ford, Carolyn 21:05 Mm hmm. Counter unmanned systems and things like that, you have to be able to give a a higher level of autonomy to a system. Otherwise you might as well not have it. Ford, Carolyn 21:27 Yeah. So I don't know if you can talk about this, but you said one of the things that you've worked on are drones. So how does can you talk about how AI comes into that project? Is it? Documenting is it actually? Teaching the drone what you want it to do. Can you talk about this?

Oh yeah. So here. What? I'd done some work in the defense world, but then what we did here at Carnegie Mellon was I worked with some of the extremely gifted engineers folks over in the robotics lab. They created a life-size thing about the size a little bit about size of a sedan. So they designed. A. Fully autonomous drone that will compete against others for a search and rescue missions. So the idea is you can give this drone a task. It's fully equipped to go out, rescue, get on scene with a person, whether they're on the land or in the water. It handles its own navigation. It can work in teams of drones.

Ford, Carolyn 22:39 No pilot. No pilot. You basically tell it the mission. So again, very R2, right? Ford, Carolyn 22:45 Yeah. You give it the mission and then it goes. And it figures out the best way to get there. It figures out the best way to find the person. Gets itself there and back with considering everything that you can figure out for it. There's a lot of artificial intelligence that plays into that. Obviously, the more times that it does the tasks, the better it gets. The more experience that you can give it. Navigating an area, the better it gets. In particular, for me as a search and rescue helicopter pilot, it's a little bit sad. That's because I'm. I'm seeing my job go away.

Ford, Carolyn 23:17 You're about to be obsolete. Hey. Oh, I think I've been exactly. Ford, Carolyn 23:22 You've been replaced. Well, OK. I I may already be obsolete. Ford, Carolyn 23:25 So now I'm thinking about all this data in this autonomous drone search and rescue drone. Mm hmm. Ford, Carolyn 23:32 A lot of it's got to be classified. You even probably need to move classified data to the drone, maybe. I don't know. How do you move it back and forth? Or do you not until the drone comes home?

Well, this kind of plays to any situation. So you talk a lot about classified data, right? And this is where you get to risk, where we were talking. We're starting to talk earlier. There's hardware solutions, there's software solutions pretty much anytime dealing classified realm that the answer, if it's not, if someone's coming to you and telling you that the answer isn't both, they're probably not the folks that you want to talk to. You need a mix of both. I mean, you've had folks on the podcast before talking about data diodes and smart proxy solutions and things. Frankly, that's not my realm of expertise. But going back to kind of the the crux, I guess of the article in a way the prompt engineering comes in because you say, hey, you need to move data between environments, right? And so the first question to ask is do I need to move data between environments, right? So again, come back to your problem.

Ford, Carolyn 24:46 Mm hmm. Start at the beginning. Hey does my user whether the user is the drone or the person operating or the person in the water, do they need that data or do they need some other version of data that would do just as well right? So. Ford, Carolyn 25:00 Mm hmm. And because we can do these things, we can move this data. We can create these tools, we kind of build this tendency again towards because the tool is there because AI is there, we should have it because we can move data, we can get more data across. From Nipponet to Cyprinet higher, we should just put it all out there and move whatever we can. But the the real it's you had General McChrystal on during one of your early episodes and he talked. I loved his the portion he brought up about inertia, right and the role it plays in risk.

Ford, Carolyn 25:37 Mm hmm. And I think in this situation you get into kind of a can we may we should we paradigm you know yes we we can move this data. Ford, Carolyn 25:46 Mm hmm. You know, are we allowed to? Should we? The real question is, is why are we moving this exactly? Ford, Carolyn 25:56 Yeah. Do we even need to? So do you build that? Do you build that kind of? I. I don't. I don't know if it's a feedback loop, but do you build those kind of questions into the AI and you tell it? Ask yourself this. All the time before you try to move something like you AI, you ask.

Right. Ford, Carolyn 26:14 I shouldn't have to ask you this. This is just part of the script. Right. Right. Well, I think what you're talking about, right? Security by design, right? Ford, Carolyn 26:22 Yeah. You're talking about, hey, start at the beginning. So if if it never needs to be on the system, we'll keep it off the system entirely, you know, and then you're gonna be absolutely secure. And then.

Ford, Carolyn 26:24 Mm hmm. Creating the system to ask, you know, to clarify for itself, hey, do I? Never ask for something, right? Basics of security. You know the the clearance and the need to know. So does the system need really need to know what it's asking for? Ford, Carolyn 26:52 Well, yeah. I mean you have just like logistical constraints to bandwidth. I mean, you know, you want to only move what you need to move and what needs to be moved.

Oh yeah. Right. Ford, Carolyn 27:07 So so implementing these solutions like it all sounds good. And like let's magically go get me some of that AI to quote you from earlier. What steps can leaders take? To ensure that AI integrations are practical and mission aligned, like let's get real. Cool. I think that's that's why in that article that I wrote, I took the time a big chunk of the article to to again my my target audience was not probably your a lot of your normal. Folks that are listening to the podcast, it was not a technical audience. It was those senior leaders who are, you know, freshmen to the subject. And my tip is my first one is really know a little bit about what you're talking about. Kind of like the Internet. And we all thought it was a fad, but it's it's it's AI. It's not going anyway. I mean, there's a lot of hype around it as we've discussed, but you really can't get away from starting to have at least a basic education about what these tools are, the differences between them, the fact that you can't just slap AI as a big label on all. Of these different things and then talk about them as if they're one.

So really step one is kind of knowing the basics and how they work. So that you know what you're asking for. And then I say the next two are really focusing on data and infrastructure, which is the some of the non sexy part of AI, but that that doesn't make it on the posters or the you know the the 10 second ads on on Facebook or LinkedIn.

Ford, Carolyn 28:41 Mm hmm. Mm hmm. For data thinking just hey again, do you really need AI or do you just need some sort of business intelligence? Do you need some sort of deep insight from the data that you have? Do you have a lot of data or do you need some sort of new? View from it that you can't get with your workforce, or do you just need your data organized and displayed for you in a way that you can make a better decision because the expense for those two things is gonna be quite different if nothing else. And then the.

Ford, Carolyn 29:17 Mm hmm. Other that we saw a lot in the defense world and we see no. Ford, Carolyn 29:22 So wait. So so, Scott, you're saying they should figure out what they need first? What? Ford, Carolyn 29:31 All right, so second one. Exactly. And the second one is it's infrastructure. So there's there's two sides to that, right. One you you need the compute you need AI these things they they run. They're very, as you mentioned before about the data transferring it, moving it, storing it, processing it. It's much more complex than things orders of magnitude more complex than things we were asking Defense Department, and, you know, normal workstation computers to do 3-4 years ago, right?

Ford, Carolyn 30:07 Mm hmm. So if if you if you aren't already investing in an infrastructure that can handle the things you're gonna ask it to do, you're setting yourself up for failure, right? Most companies will explain that to you once you're there dealing with them. But if you're on the other side of the contracting, if you're just the idea guy in a uniform. That's a great way to get yourself in trouble, not understanding it. Plus the other side is culturally. We talked about adoption before. Or, but if you're gonna go to your workforce and you're gonna say, hey, I want you to. I've got you these new tools. I've got you this new AI tech and they're living in a world where their printer doesn't work. They can't stay logged in. You know, their e-mail lags. You know, they're waiting for things to refresh. You already bred some cynicism in there, and you're gonna be finding an uphill battle with adoption.

Ford, Carolyn 30:59 Yeah. Well, and when what you just said just the the compute that's necessary. In defense, I would think that you can't just go get you some AI. The way I can go get me some AI when I'm talking about. I'm talking about like generative AI. Good. Ford, Carolyn 31:16 That's what I'm using everyday to help me with my tasks and. Mm hmm. Ford, Carolyn 31:22 In defense, you're not just gonna connect to the Internet and start putting your information and data. Out there like you, you have to be able to contain it yourself. And so you have to be able to manage that compute in your own infrastructure. Like I'm using somebody else's infrastructure, right?

Right. Right. Right, right. Ford, Carolyn 31:42 So yeah. All right. Well, what are some measurable outcomes or changes that you've observed? When leaders can talk about. AI in an informed way and understand like the different types of AI, quite frankly from GPT, which like I said is generally my world. AI that's running your infrastructure. Back end. Whatever it is, I don't really know 'cause I don't use that. But but what? Like what have you observed? When leaders can talk about it intelligently. Within military and defense.

Well, I doubt it's unique to the military or different school. But before coming here, I was fortunate to work in a capability development branch. And so we would deal with the creation of tools from both the commercial sector, bringing them in and then internally with the team, some extremely smart people that we create some of our own or or other, you know, government to government tools that were built that we could. Integrate. Into the work that the people at our command were doing. And a lot of those tools, we spend a lot of money, we'd spend a lot of time, we'd bring them in over months and months and months, you know, iterate on what was being created and then kind of roll them out. A lot of the ways. To some of the the analysts, the folks that would be using them and we, you know changes change is scary, right? So sometimes we hit on a lot of, you know, that's cool. But I don't really want to take the time to learn how it works. Learn how to use it. We hit some adoption issues, right? But then we would find, I think we referred to them as analog problems because they were these small things that got in people's way in their day-to-day.

And we'd be like, well, yeah, that's easy. I could give that to to the one of the folks on the team in three days. They could create a solution, you know, pretty low tech, just rules based stuff that really fix their problems that could solve it. And and they adopted it.

Ford, Carolyn 33:53 Mm hmm. Right out the gate and the the thing that comes to mind when you ask a question is. We would start to see social networks within the command that we could kinda map because they would. They'd see the tool that we created and then somebody that they tell their friend and then their friend would come down to us and say, hey, I heard you help. Ford, Carolyn 34:10 Hmm. So and so actually solve their problem. Can you help me with mine? And then we'd help them. And then you know and then we would see folks say, hey, come down to our shop, come down to talk to people and you'd be able to tell who told who, you know, was watching.

Ford, Carolyn 34:26 Mm hmm. Is like watching a good rumor spread and it was fun watching. You know, when we weren't advertising and we weren't trying to push something out. Ford, Carolyn 34:29 Yeah. The. The utility of what we were putting out there once folks got their hands on it and once they were able to work with us to create it and they understood that one, there was a level of trust there, right? They weren't. They knew we weren't selling them something and we were just there to help and we were actually going to solve their problem. Or if we couldn't solve their problem, we'd tell them.

Ford, Carolyn 34:50 Mm hmm. And so you could see that network of people feed in and we could and then we could use that as a lesson and say, hey, whenever we have these big issues, hey. By by getting people in on. In the creation process, right that that that's how you kind of see those wins really play out. Ford, Carolyn 35:15 So like you said at the beginning of the show, using foundational. Human skills like empathy, identifying challenges and then finding. Creative, innovative, effective ways to solve them. And I mean, as you're talking about it, I'm like just build me an agent, Scott, and have it come in and solve all those problems for me. I don't want to learn how to use it. I don't even wanna know you're doing it. Just fix it.

Mm hmm. Yes, and I'll and I enter the user's point like the leader needs to be familiar without their app with what they're asking for. Ford, Carolyn 35:49 Yeah. But the user the user really they only need to know enough to be useful to be able to use the tool, right? I don't need to design it and so and that's the role you should play in the process if if that's what you're asking for. Your role really extends to the point of making sure that I understand what you need.

Ford, Carolyn 36:06 Yeah. And then at that point, you can just tell me. Yeah, it works. No it doesn't. Ford, Carolyn 36:23 All right. Well, we're coming to our fun, rapid fire tech talk questions. OK. Ford, Carolyn 36:31 So if you could delegate one of your past Coast Guard duties to AI, which would it be? 2. E-mail. Ford, Carolyn 36:40 Right. No, I mean that. Ford, Carolyn 36:42 I don't know. That one's universal, right? Oh, yeah.

Ford, Carolyn 36:44 It could be expense reports for me. I mean, I I say that one again, almost tongue in cheek. But I mean, who doesn't want to get rid of those? Those things and the and then at the same time, I'd say the one that I don't want to, but I probably should. And I would say if I stay around long enough, will is search. Ford, Carolyn 36:54 Yes. And search and rescue right? There's, there's almost no reason to me.

Ford, Carolyn 37:07 Oh. In the near future, for us to be sending, you know, putting people especially domestically in harm's way, to go look for someone that needs to be found, if we if we need to be there to rescue but. Ford, Carolyn 37:18 Mm hmm. The tools exists. The autonomy exists. You know the the AI in the robotics can work together. They can. They can get on scene faster than me. They can search better than me and they're hopefully hopefully a bit more attitable than me.

Ford, Carolyn 37:34 Wow. Yeah. So I I I think I'm soon to be. I should be soon out of the job. At some point in in that sense. Ford, Carolyn 37:45 Just stay at Carnegie. Just stay in that think tank. Yeah. All right. Well, imagine dropping today's AI into your childhood, the 1980s. Mm hmm. Ford, Carolyn 37:56 Into a military operation. What's the first thing that would change? You know, before coming to Carnegie Mellon here, I think my answer would have been really more informed by things like short circuit from the 1980s and, you know, been in Johnny 5 situation and it it would have been you know, 3 Stooges impersonations and it would have.

Ford, Carolyn 38:16 Yeah. Been a lot of fun, but I think. Audrey Kurth Cronin. She runs the CMS program here and she's done a great job of highlighting. Ford, Carolyn 38:29 Mm hmm. The unintended consequences? That can come with these big technological jumps, right when you look at history. Anything from dynamite to what are some of the things she has the AK47 social media, exactly right. Things that we thought we could.

Ford, Carolyn 38:45 Plastic. Yeah. Only some things we thought we could only see the good side of, and then someone found a way. Or they just evolved themselves into an unattended consequence that we didn't prepare ourselves for, I think. The thing that we, the one thing that we have going for us is that we're evolving. With the technology as it goes, the the pace, particularly with AI is obviously so much faster than other things in the past. And I think that's what scares a lot of people kinda moving from.

Ford, Carolyn 39:15 Mm. We've moving from playing chess to playing speed chess in a lot of ways, and so I think now that I've been here. Ford, Carolyn 39:23 Mm hmm. I I think it would pretty much. I'd be worried for society as a whole if you know, I think the results could be pretty catastrophic if you know 1 military in the 1980s suddenly had what we have now. I don't know if we'd be able to adapt quickly enough.

Ford, Carolyn 39:39 Yeah, right. It's just like, let's not. We're just not going to do that. We're not. Exactly. Ford, Carolyn 39:44 We're not going to send Terminator back. We're not doing that, all right. We saw how it played out. It was not good. Ford, Carolyn 39:50 That's right, if AI were a recruit, what would its boot camp training focus on? Mm hmm. Probably not swimming. Ford, Carolyn 39:59 OK.

No, not I think. I think you you focus on what is good for. So I think it would focus on the service history like particularly to our own, you know, Coast Guard history is something that it would it need. Probably learn a lot quicker. I know it would learn a lot quicker than we do. But I mean think about it. It can learn. We learn lessons from past generations, right? And so we, we're happy when our leaders can really take those things and they can avoid making the same mistakes that we made, you know, one more ago.

Ford, Carolyn 40:33 Mm H. You know, they always say you're fighting the last. Don't fight the last war. But we only have so much capacity for that context, right we can. Only we can only really understand or know so much. Looking backwards at our our nation's history or society's history. Whatever. But AI has the opportunity for full scale context to look at lessons. Ford, Carolyn 40:56 Yeah, limitless. It feels like limitless, yeah.

Exactly right. So where you know, my leaders may consider the last 30-40 years if they're. If they're great, you know, I would love to have an AI that came out of boot camp that could really give me a decision based on the totality of history that I'd understand. Ford, Carolyn 41:00 Yeah. Mm hmm. Well, thank you so much for talking to me today, Scott. It's been really fun, even though my head was a little broken. Like I said, I was just like, OK.

Oh no thank you. Ford, Carolyn 41:26 I misinterpreted your article cuz I read what I wanted to see right. And so just talking to you, I'm like, oh, you just expanded my mind, which I love. So for leaders aiming to integrate AI into their operations effectively. Let's. Sift back through our conversation. Give me one actionable step that they can take today to start applying these principles that we talked about.

It probably would be redundant, but I'd say spend some time redefining the problems that you think you're solving, and do that with your with your workforce. Don't do that by yourself. Ford, Carolyn 42:13 So are we redefining? So we really understand what the problem is trying to get to the heart of the problem. Mm hmm. Ford, Carolyn 42:20 Yeah. OK. I like it. All right. Where can our listeners connect with you to learn more about your work and your insights?

Here at CMS for another couple of months, but just like anybody else I can give you my LinkedIn or e-mail address if people want to reach out. Ford, Carolyn 42:38 I will drop your LinkedIn into the show notes, so thank you again for joining us. Thanks listeners for tuning in. Please share this episode, smash that like Button and we will be back soon with more from the world of Tech transforms. Stay curious and keep imagining the future.

Thanks again. Ford, Carolyn 43:00 All right. That. Ford, Carolyn stopped transcription