ICYMI: Enabling the Workforce to Capitalize on the AI Opportunity

This episode features a panel discussion on the role of AI in workforce development within federal agencies at the AI Acquisition Forum 2025. Led by David Vennergrund, Vice President for AI at GDIT, the panel includes distinguished guests: Andrea Brandon from the US Department of the Interior, Florence Kasule from the Department of Education’s Federal Student Aid office, and Wole Moses, Chief AI Officer at Microsoft’s federal civilian division. The conversation delves into the evolving landscape of AI, focusing on training, innovation, and integrating AI across various federal roles. They also discuss practical strategies for implementing AI, such as sandbox environments and community-driven initiatives.

Subscribe on your favorite podcast platform to never miss an episode! For more from ACT-IAC, follow us on LinkedIn or visit http://www.actiac.org.

Learn more about membership at https://www.actiac.org/join.
Donate to ACT-IAC at https://actiac.org/donate.

Intro/Outro Music: See a Brighter Day/Gloria Tells
Courtesy of Epidemic Sound

(Episodes 1-159: Intro/Outro Music: Focal Point/Young Community
Courtesy of Epidemic Sound)

Chapters

• 2:00 Panelist Self-Introductions and AI Passions
• 6:37 Workforce Development and AI Adoption
• 13:22 Effective Practices for AI Workforce Readiness
• 21:59 Policies and Investments for AI Readiness
• 31:48 Recruiting, Retaining, and Upskilling Workforce

Transcript

SORAYA CORREA: [00:00:00] Up to to, to introduce our next panel, our, I think it's our, our last panel and then we'll have a keynote address is David VENNERGUND . He is the Vice President for AI at GDIT. So David, if you don't mind joining us up on the stage. Yes. And I'd also like to invite your panelists to come on up if they're ready.
SORAYA CORREA: So we have, uh, I don't know if Andrea's here, Andrea, Brandon, deputy Assistant Secretary, US Department of the Interior. Florence KASULE, I hope I pronounced that correctly. Deputy Executive Director, FSA, contracts at US Department of Education, federal Student Aid area. Uh, WOLE Moses WOLE. WOLE. WOLE. I knew I was gonna get there somewhere.
SORAYA CORREA: Uh, WOLE
SORAYA CORREA: Moses, chief AI Officer at Micro. And who am I missing? Am I missing someone? That's it. We're, no, I think I got it. I got everybody. So take it away. Alright, 
DAVID VENNERGRUND: uh, testing. Good morning or good afternoon. Thank you for sticking around for our exciting panel on workforce development and my good old friend, Tim.
DAVID VENNERGRUND: Good, good to [00:01:00] see you, Tim. All right. So if you'll excuse me, I wanted to intro let each of our, um, panelists, uh, introduce themselves. And maybe share a passion or two about AI and the workforce. Um, I'll go first really quickly and uh, uh, as you heard, I lead AI at the GDIT Center of Excellence, hands-on data scientists.
DAVID VENNERGRUND: Been passionate about AI my entire career, and I'm really happy to see the revolution. We're living in technological revolution we're living in now 10 years ago with deep learning, two years ago with generative ai. And as we move forward, we're automating. And accelerating the things that we can do. So super excited to be here today.
DAVID VENNERGRUND: With that, let me, uh, turn it over to our panel, Andrea. 
ANDREA BRANDON: Hi, uh, Andrea Brandon from the Department of the Interior, deputy Assistant Secretary for budget, finance, grants and acquisitions, and also property and small business and suspension department, and also a wonderful group called the Business Integration Office, uh, the bio.
ANDREA BRANDON: Um, and [00:02:00] they are responsible for all of our, uh, system integration system, financial system, grant system. I ran, uh, acquisition system, our RPA robotics process automation and our artificial intelligence, uh, integration efforts. And basically my passion for, uh, all disruptive technology kind of started back in 2016, um, when I was in another position at a different federal agency.
ANDREA BRANDON: But it's, it's really a vision that, um. Compels me for all the federal government with regard to artificial intelligence. And, and it really stemmed from, um, watching the movie iRobot and there was an, yeah, an artificial intelligence with Positronic brain called Vicki. So I like to tell people I want Vicki, and even though Vicky kind of scares people a little bit, you know, I always tell them, we'll figure out a way to shut it down just like they did, but if, if necessary.
ANDREA BRANDON: But yeah, I really am a nerd. With regard to all disruptive technologies, I could talk all day about it, but I'm gonna pass it on to our next panelist. Yes. Thank 
FLORENCE KASULE: you. 
DAVID VENNERGRUND: And hopefully Hal and [00:03:00] Vicki don't get 
FLORENCE KASULE: together, 
ANDREA BRANDON: know exactly. 
FLORENCE KASULE: Hi everyone. Uh, my name is Florence Cast. I am Deputy Executive Executive Director for FSA Federal Student Aid, which is within the Department of Education.
FLORENCE KASULE: I have to caveat and say the thoughts and opinions that I'm sharing today do not represent the my agency. I am here on my personal, um, behalf. But, um, here to share observations that I've had, um, across different agencies because I've been at a number, most recently before Department of Education, I was with, um, a little organization called the US Digital Service, um, and have led, um, I've been in the acquisition space for a few decades now.
FLORENCE KASULE: I can't believe I can say that. Um, um, mostly working on the. Acquisition of technology and enabling, um, different missions. Um, and why I am excited, my passion is really using technology, the combination of what we are doing with as federal employees, with [00:04:00] our, um, industry partners to make things better. To make experiences for people better across, um, the us um, and abroad.
FLORENCE KASULE: And, um, yeah, it's, it's, and just using, using technology to make things better for people and really within our, this, this session here, making sure that the workforce is empowered, um, to use technology safely. Cool. Rolling. 
WOLE MOSES: Thank you and, uh, thank you to act, uh, act I act for having me here. So I'm Wole Moses.
WOLE MOSES: I'm the Chief AI Officer for Microsoft's federal civilian division. Uh, I'd say my passion for AI started when I was maybe seven or eight years old when I used to build robot robots from, uh, robotics kits from a store called Radio Shack. And this is an age, age test. If you, if you're, if you're of a certain age, you'll know what I'm talking about.
WOLE MOSES: If you're, if you. If you're of a different age bracket, you'll have no idea what Radio Shack is, and so. [00:05:00] That's where it started for me. And then similarly to, uh, similar to Andrea, uh, it, it mo it went to, uh, Hollywood with the Terminator, and I was like, is this thing real? And so, um, and then that followed me to, uh, what I ended up studying in college and then into my career at Microsoft.
WOLE MOSES: I've been at Microsoft for 26 years, which I know is a long time to be with one company, but I've had an incredibly diverse career, and right now I'm, I'm excited about. The space that I, that I lead in terms of our AI strategy at Microsoft. And so excited about the panel today and, and, and hearing the questions.
WOLE MOSES: Awesome. Thank you Willie. 
DAVID VENNERGRUND: Alright, well I'm gonna set up the first question, uh, by observing, there are two kinds of people in the world, those that group things into two groups and those that don't. So here we go. The groups that I, uh, see are users of AI and builders of ai. And if we'd been here five or six years ago, we would've been talking about [00:06:00] Python and deep learning and CNN and all the great technical stuff that we revel in.
DAVID VENNERGRUND: But we're not, we're, we're not, we're not focused on builders today. We're focused on users because the power and strength of AI tools has blossomed so greatly. Everyone has an opportunity to use ai, um, and. As Florence pointed out, they're using it at home. So, uh, we've got gorilla workers trying to use it at work, so let's, how do we get control of that?
DAVID VENNERGRUND: How do we make it safe? What are some ideas or ways that your organization, Andrea, are, um, effective strategies that you are using to prepare the workforce? To adopt and use AI in everyday tasks, not the building tasks. 
ANDREA BRANDON: So I'll, I'll start by telling you a quick story. Um, about a year and a half ago, the Department of Interior rolled out DOI chat, GPT.
ANDREA BRANDON: And you know, we were, some of us gorilla people were like, oh, this is great, you know, finally we've got it, you know, rolled out [00:07:00] across the entire department. But immediately I am as a das. Started getting emails from my staff, all levels of staff, and they were telling me, what is this? It's just a black screen.
ANDREA BRANDON: I can't, it's not working. You know, I've put in, I've asked it questions, it's not working. I don't think it has the right data, et cetera. Now I knew that it was formulated from regular, uh. Um, the regular chat, GPT open AI's regular chat. GPT probably had, you know, the 2022 data and you know that whatever the data models that they used to train it.
ANDREA BRANDON: But, so I knew it had data and it had information. Um, but I pretty much figured from all the emails I immediately started getting that they didn't quite know how to use it. Mm-hmm. And so therefore they were going back to Google. They're like, this is worthless. You know? So basically I created a prompt sheet, um, that had different, it had 10 different prompts on it, but some of the prompts were related to financial management.
ANDREA BRANDON: Some were related to contracts, some were, uh, related to grants. Some were related to budget. [00:08:00] Because I tried to get a flavor, give a flavor, like, you, you know, if you work in this area, you can use it for, you know, put this prompt in and then you can kind of get a gist of how it works, and then maybe it would kick them off.
ANDREA BRANDON: Um, into doing that. So I think that that has helped. But also DOI has a community of practice, um, where, and it's all across the department and you can join it. Um, and we send lots of messages through the team's chat, but there's also training opportunities for AI through the community of practice. We have DOI University, we've had several training.
ANDREA BRANDON: Classes through DOI University. And then one of my favorite things to do, which we can talk a little more in depth about, uh, later in one of your other questions is, um, I like to drag my team to innovation labs. So we've gone, uh, probably three or four times to the Microsoft Innovation Lab. You know, we've gone to SAP up in Hudson Yards, you know, because we use SAP financials.
ANDREA BRANDON: So, I mean, we've seen some really great stuff there and in the other innovation labs we've gone to as well. So that's a really good [00:09:00] way to have people do a hands-on. You know, training, learning, you know, touch it, feel it, actually spend a day working in it, and then they can go, it's like a train the trainer approach.
ANDREA BRANDON: They can come back and train other people. 
DAVID VENNERGRUND: I love it. So just teasing that out a little bit, you have prompt sharing. So you have groups that get to you, you write the best prompt, share them, and they get passed around kind of like, uh, maybe recipes. Yep. You have a community around recipes. Yeah. Yeah. You, you've built a community.
DAVID VENNERGRUND: People can interact. You have training, uh, venues and you have innovation at venues. Yes. Great. Florence, would you like to add to that? How, how is education looking at building up the workforce? 
FLORENCE KASULE: Sure. Um, I think I, I can expand on beyond education, right? Like there are, um, every five or so years, um, we've had some type of technology that's jumped in on the scene, right?
FLORENCE KASULE: That everyone is circling around. Um, and. The way that I've seen it be successfully implemented [00:10:00] is, um, across different agencies, because when I was with USDS, we, our footprint touched a lot of different agencies. Um, is when people would identify power users, those first movers in different organizations.
FLORENCE KASULE: I'm an 1102 by training. Um, and, but my 1102 hat also extends to my business unit hat, right? I wanna make sure that I understand what are my. Business units, what are the, the different organizations, the requirements developers, what are they interested in? Um, but that's across all organizations. So identifying those power users, the first movers who, who are interested in whatever that technology is, and inviting them to sessions to learn about it, understand about it so that they can proselytize and say, and kind of do the, someone on one of the former, um, sessions was talking about starting small.
FLORENCE KASULE: Building momentum and I strongly believe in that. And getting, if you get that core group of people [00:11:00] together who can learn and say, and send you the emails to say, Hey, this isn't working. And then when it does work, say it works now and, and spread the news to people, right? Um, but getting that core group of people together from the different organizations.
FLORENCE KASULE: So rolling out new technology like AI is not contained only in the CIO shop. It's not only in the CTO shops, but that it is touching everyone and everyone sees themselves at the table. 
DAVID VENNERGRUND: Great. Um, I heard at a recent conference, if you want to speed the adoption of AI across your organization, start at the top, get your CEO, um, off to the side and do a chat session with them.
DAVID VENNERGRUND: Do you, do you feel like that is sort of the model you're using there? 
FLORENCE KASULE: Uh, not 
DAVID VENNERGRUND: quite. 
FLORENCE KASULE: I think you have to, yes. Um, ownership comes from the top in terms of making, um, strategic decisions of what is going to come into the environment. But you really do need a [00:12:00] groundswell. You need, it's introducing and enabling technology can live and or die if you don't get a proper groundswell, right?
FLORENCE KASULE: If you don't get adoption and people, you can buy, I in procurement, we can buy amazing things, but if people are not willing to use it. They can choose not to use it optimally. 
DAVID VENNERGRUND: Cool. Thank you for that, uh, warning. Appreciate that. Um, I'm gonna shift over a little bit to innovation. Um, if I could, you're the, you're with an industry leader, Microsoft, you're leading with open AI and Azure Services.
DAVID VENNERGRUND: Um, you work with a lot of public sector organizations. What have you seen as some of the more effective practices for agencies when they're investing in workforce readiness in ai? 
WOLE MOSES: Yeah, so it's a great question. And so to your point, like we sit in a seat where we, from a Microsoft perspective are enablers and so.
WOLE MOSES: We work across the federal landscape. The federal agencies are our [00:13:00] customers, and we work with them around, uh, learning about ai, experimentation, proofs of concepts and MVPs, all the way to technology deployments. And so I'd say a lot of what's already been touched on has been, has been critical. And so in terms of workforce development, one of the, one of the things that we've seen is, you know.
WOLE MOSES: AI is, AI is new in particular from a generative AI perspective, and so there, there's a few things that are different about generative AI that it's important to make sure everyone understands, and so basic AI literacy. Is important. And so the fact that generative AI as opposed to traditional AI that we've experienced over the years, generative AI is probabilistic.
WOLE MOSES: And so you need to check responses. I think the previous panelists from DOE talked about that. And so, uh, you need to make sure there are references cited to some of the outputs in, in that, uh, there's a human in the loop. And so that's one part of it. But the other place where I think it really, [00:14:00] that we've seen it become really powerful is that.
WOLE MOSES: When organizations invest, not only just in in in basic AI literacy, but in role specific AI use cases. And so that's where we really see momentum being driven. People understand. Okay, I understand sort of these general use cases, but in my role in procurement or in my role in hr, in my role in research, how can I use these tools specifically to accelerate the work that I do?
WOLE MOSES: And so that is one of the best practices that I'd call out that we've seen across the landscape of customers we work with. 
DAVID VENNERGRUND: Excellent. Thank you. Um, at GDIT, we have a whole series generative AI for. X for software developers, for data scientists, for communications, for legal, for hr, and all those are built on, um, learning courses.
DAVID VENNERGRUND: We can get through different vehicles and often pointing right back to Microsoft and other. Providers of those services. So differentiation really does, really helps people see themselves in [00:15:00] ai. That's, let's pull some more on that innovation thread. You, um, we were, we were talking, uh, before this session about your team's experi, well you mentioned it today, experimenting, uh, I didn't, you didn't use the word sandbox, but how do you bring innovation?
DAVID VENNERGRUND: How do you, uh, allow people to experiment? Try things in a, uh, risk-free environment. What are some approaches you take? 
ANDREA BRANDON: So, um, basically we do have somewhat of a, an AI sandbox within DOI, but also what we do is we allow, from the ground up, they can provide business cases, use cases. We do collect them, and then we do like an analysis of which ones seem maybe a low hanging fruit, if you will, less risky, um, that we can actually work with whichever vendor.
ANDREA BRANDON: Um. That has the AI that we're interested in. It might be Microsoft. Sometimes it's dealing with a copilot project and sometimes it might be just a general AI project. But yes, we do, um, collect the use cases. We look at the low hanging [00:16:00] fruit ones, the ones that may not be as costly to try to implement the ones that are less riskier, less.
ANDREA BRANDON: That may cause less fear. Um, and that would be very beneficial and for efficiency of whatever the line of business is, that that has brought up that use case. And then we allow them to pilot the, the use case. Um, we build it, um, and then we look at the results from the pilot. Then we've had a few that are gone into actual production because they have been very successful.
ANDREA BRANDON: Of course, we do have to look at, um, is the information real or is it memex? And you probably remember that if you're older. But, uh, just making sure that it's given us reliable information and, uh, as we build out efficiencies and so forth. Um, I think also it's good to get it, the information from the bottom up because a lot of people were fearful.
ANDREA BRANDON: That AI was gonna replace them. And so now they're using it. They're, they're participating in the pilots, they're coming up [00:17:00] with the use cases, and they're bringing it up more and more. And our use case list is actually growing. And, but that didn't come from the top, you know, from the upper leadership. The use cases are coming from the bottom and from the middle level, from, from the middle tier of staffers.
ANDREA BRANDON: So definitely, yes, we have like a, we do have somewhat of a sandbox that allows us to safely pilot the ai. This. 
DAVID VENNERGRUND: So I know you emphasized low risk, but it feels to me like you're encouraging them to take some risks by proposing use cases that perhaps no one would ever have thought of before. So, um, Wolley, I know that you've got, um, an ability to try to experiment with things, if you will, to dialogue, right?
DAVID VENNERGRUND: Here's the use case, but also here's a capability and we need to be able to say, what could I do with that capability as well as how might it, what could it satisfy? So tell us a little more about your innovation, um, uh, uh, incentives or labs or capabilities. 
WOLE MOSES: Well, yeah. So first, uh, shout out to Andrea. Uh, so do OI is [00:18:00] one of the federal agencies that was one of the first to lean into and adopt generative ai.
WOLE MOSES: So we've worked very closely. We were trying to remember if we've been in the same meeting space at our, at our Rozen office. I think undoubtedly we, we have, uh, so yeah, I mean, we have a, uh, facility where we, we help customers come in and, and, and envision and ideate. We do ideation sessions. Where we talk about, from a Microsoft perspective, we talk about what's possible with the technology, and then from an agency perspective, the agency brings potential.
WOLE MOSES: Uh, problems or, or challenges that they're looking to solve. And then we, we sort of figure out together whether or not AI is the right solution. In some cases it's not. In some, in some cases it is, but it's a, it's a good exercise. One thing I wanted to touch on is, is the whole conversation around risk and a identification of low risk use cases.
WOLE MOSES: That's been a, a topic and a question and a concern, um, since the beginning of, uh. Folks looking at generative ai, [00:19:00] one of the things that we have an opportunity to do just given our footprint and our engagement across the federal agency agency landscape is, is many in, in many cases we've seen that agencies are embarking on or looking at some of the same AI use cases, are solving some of the same problems, but in isolation.
WOLE MOSES: And so one of the things that we like to do is facilitate agency to agency conversations where. One agency has identified low risk use cases. We make those introductions. Sometimes we're involved, sometimes we're not in the subsequent conversations. And that's one of the ways that we've seen, uh, a healthy, uh, movement forward and application of, um, of these conversations that in fact, I've, uh, with the grant work that DOI has done around the, the grant proofing, uh, solution, we've introduced several.
WOLE MOSES: Other agencies too, the folks at DOI who led on that initiative. So that's just one example. 
DAVID VENNERGRUND: Excellent. Yeah, I think the publication and of, of [00:20:00] inventories and sharing success stories is critical to that diffusion that you described. Florence. I'm gonna ask, go back to you for a moment. Um, looking ahead, what kind of policies or investments do you think are critical to ensure the federal workforce is ready for ai?
FLORENCE KASULE: Okay, so investment number one. Is training, um, I know a lot of budgets have been impacted, um, but that doesn't mean that you can't get creative and do internal training or like will they, um, just described this, what I call phone a friend is go reaching out and saying, Hey Andrea, what are you doing at your agency?
FLORENCE KASULE: What, what has been successful for you? How have you trained your folks? Um, and doing, building on the building your internal muscle, um, and. Or, um, procuring training as well, that can, that can, that's also an option, but training is number one. Um, just as a show of hands, how many people here have two phones with them [00:21:00] right now?
FLORENCE KASULE: Some of you have three or four. I used to live that life as well. Um, but on one of your phones, likely you might have some AI tool. Right. Um, for many of us in the government we have, we're working with two, three, et cetera phones, and there are many of your government counterparts who may not have an AI tool on their work phone, right?
FLORENCE KASULE: So you have a lot of people who are operating with their personal, in their personal lives on AI tools, very comfortable doing a lot of things. With the recipes coming up with recipes for home or summer camps or vacations, whatever, and it's, it's benefiting them in their personal life. But then they jump into their professional life and they, they don't have that same type of fluid.
FLORENCE KASULE: Benefit from ai. Right. Um, and so I think the inve, one of the investment pieces that I think, um, is important is [00:22:00] understanding the diversity of your, of your workforce and what people are using and be, and, and leaning on in their personal lives. And how can you, um, provide tools within their professional lives to help boost and, um, improve how they are being efficient.
FLORENCE KASULE: Right. Um, but in terms of policies around that, um, training is one, figuring out how do you make sure that people are AI literate and know what that means. So everyone is, every day AI is on some front page of something. Um, and I wanna make sure that the workforce that I'm working with, I'm particular to the, um, procurement world, that they know what that is.
FLORENCE KASULE: Because sooner or later, if it hasn't happened already, you're gonna, A package is gonna hit somebody's desk or inbox to say, you are, you are going to procure some AI something. Um, what does that, I don't want that person to, when it hits their [00:23:00] inbox to say, what does this mean? It's too late at that point.
FLORENCE KASULE: I've always said, and I strongly believe, if you don't know what it is, you do not know how to procure it well. And so I wanna make sure that the workforce, particularly the procurement workforce, understands the, um, complexity, the nuances, um, and the many, many benefits, as well as the risks associated with buying all types of the flavors of AI that are out there.
FLORENCE KASULE: Um, part of that is something that Wole mentioned, um, is keeping the human in the loop. When I've heard of, um, there are agencies that are using procurement tools that ha, that are AI driven, um, that are helping craft statements of work, helping craft, you know, sos, et cetera. Um, making sure that, that there's, when something is generated, that there is someone actually reviewing it.
FLORENCE KASULE: And, and understanding what is it putting out there? If it's a solicitation, is it putting out proper? Is it [00:24:00] actually creating proper, um, the proper information? Um, I'll stop there. 
DAVID VENNERGRUND: No, that's terrific. So I heard, uh, training, which is wonderful, but I also heard sort of a, an, um, allusion to some form of a chat, GT chat, uh, service perhaps internal, perhaps, uh, secured, perhaps.
DAVID VENNERGRUND: Um. Brought down a model brought down from the internet and put behind a security wall like you may have done, right? So teaching people how to use those. But training is important, but you need those first, right? Either access to the commercial SaaS version or a local version. So you, you kind of started the story of people using one of yours and asking questions that didn't work out so well.
ANDREA BRANDON: Yeah. Um, well, 
DAVID VENNERGRUND: so one, one of the questions we get is I have this model from 2022 and it didn't know who won the Nats game yesterday. Well, first of all, they lost 'cause they always do. But second of all, you're using it the wrong way. So in addition to training, what other investments do you see that we [00:25:00] need to make?
ANDREA BRANDON: Um, so definitely, um, we need to, well, let me speak on something really quick. Sure. As she was talking, um, I was thinking, and you brought it up initially in your introduction. Where there's two types of people with ai, so I think one of the investments that we need to focus on is widening who we think needs to be involved in ai.
ANDREA BRANDON: And I say that because a lot of the policies, executive orders, regulations laws are focusing on the chief information officer, the chief technical officer, the chief data officer, the chief AI officer. But it's not just them. Who needs to use it or develop the product or, you know, buy the product or understand ai.
ANDREA BRANDON: It's all of the lines of business and it's the program and mission people as well. So when you talk about training or when you, um, general information training on artificial intelligence, we need to reach out to our mission people, to the program people and [00:26:00] DOI, that would be the people in the parks.
ANDREA BRANDON: That's the people that are on the oil rigs, the people that are in the dams, the people, because AI has. The capability of really making efficiencies and doing some better predictive analysis on the data that's coming in, on the rigs, you know, or um, uh, you know, in our fire season and so forth. So I think that we need to broaden who is AI supposed to really benefit and who should some of these regs or policy?
ANDREA BRANDON: Target. And so I think that's a very key thing. So it's, it's training, yes, but it's also broadening the target area. The target population. It's the whole government. You know, when I was at other departments, you have the scientists that are at. A specific department that I don't, I don't want to go into anybody else's, uh, lane, but you have researchers that are working in the federal government, they need to have the benefit of, of ai, artificial intelligence.
ANDREA BRANDON: And there are different types. There's, you know, yes, there's chat GPT or Open ai, but there's Dolly, there's like all different types of artificial intelligence. So I [00:27:00] don't think that. The, the field of who needs to be involved with AI is broad enough, uh, when you're looking at the, the executive orders that are being put out and so forth.
ANDREA BRANDON: And definitely, I don't think that the use case of what types of AI is broad enough. Um, I think that we're all stuck on, you know, no offense, but Microsoft copilot or, or we're stuck on, you know, China, GPT or whatever. But there's, there's so many out there. That are, you know, pre-trained for art or pre-trained for other, uh, businesses, other types of lineup, legal, you mentioned one of them.
ANDREA BRANDON: There's legal ai. We could use that, you know, we, using the solicitors and the general counsel offices, we can, so they could benefit from that. We don't have to start from scratch, but I just think that the field is limited when the policies and the executive orders and the regs and the, and the laws come out.
DAVID VENNERGRUND: I think, um, what will break those doors down to the specificity is domain specific knowledge, bringing in your rules and [00:28:00] regulations, the firefighters, um, guidelines, bringing in anything that has to do with applications or grants. Once you have domain specific knowledge combined with these powerful LLMs, then you can ask really useful and differentiated questions.
DAVID VENNERGRUND: And you can, as you point out, you have so many different roles, not just leadership roles, but from, uh, Rangers to yes. Interior, uh, business leaders, right? 
SORAYA CORREA: Yes. 
DAVID VENNERGRUND: Is there any op, is there any use case you can imagine where, uh, AI wouldn't, uh, en enable the workforce or improve the workforce? Can we think of, is there anything you would leave off the table or would you say it's universal?
DAVID VENNERGRUND: I'll start with you. 
ANDREA BRANDON: I'm trying to think. I, I don't think then it's necessary that you necess that you leave something off the table. I think that you need to derive the risk levels of the different functions that we have within the agency. And so there, if it's a more riskier function, then you need more checks and balances on [00:29:00] whatever that AI is doing.
ANDREA BRANDON: Yes. Guardrails. You need to understand that the AI should be making recommendations and not decisions, and one of the things I'd like to. Um, harp on is a, it is not just a human in the loop, but I think us as, um, as a country and other working with other countries, we need to determine what types of functions should be considered inherently human period.
ANDREA BRANDON: And, and no AI will ever do this function. It's an inherently human function, and we will always be the human in the loop. It's, we're gonna be the decision maker. It, you know, we are gonna monitor whatever those, those inherently human functions are. We need to identify them now. Not like after we have a Vicky issue.
ANDREA BRANDON: Yeah, 
DAVID VENNERGRUND: I think there's, there's, yeah. Vicky and hell could get us, but yeah. Differentiation between automation and assistance is important there. Yes. Thanks. Alright, last question and then we'll, maybe we have time for one, uh, question or so. Um, recruiting and retaining and upskilling. Do you wanna take a [00:30:00] poke at that?
DAVID VENNERGRUND: Florence? How does your organization build a workforce? Um, knowing that, uh, we have, uh, workers who, uh, who um, are close to the mission, close to the data, but maybe need AI skills and or recruiting new people. Do you have any thoughts on that? 
FLORENCE KASULE: Sure. Um, two. There are a number of points, really excellent points that, that have been made here.
FLORENCE KASULE: Um, I think we need to make sure that we have, um, we create environments where people want to be, right? So, um, and, and there are in across, I've been at, gosh, six or seven agencies in my career, and as I was, we were chatting before this session, depending on which agency you are, you're in, you find yourself in.
FLORENCE KASULE: You can, you can think that you're in 2025. You can be in 2020, you can be in 1998, all depending on the technical, the tech stack that is there. And [00:31:00] even more important than the tech stack is like the culture of innovation within that organization. And so in terms of recruiting, I would want to, um, implore anybody who's in leadership here or in HR or.
FLORENCE KASULE: Um, across the spectrum to make sure that you are creating an environment where people want to be. And, um, that includes the tools because I've had, I certainly have had people leave, um, an organization and they, and they jump into a new organization and say, wait, you guys don't use this. Like, but we're all federal agencies like you, they, I used to.
FLORENCE KASULE: And so, you know, and I'm like, I know, I get it. I get it. We're all within the federal ecosystem, but this agency treats technology differently. And so just making sure that we are having those conversations with one another as agencies to say, how did you get that over there? How are you figuring that part out?
FLORENCE KASULE: Um, every [00:32:00] one of our agencies is dealing with a, a variety of risks, right? Depending on who our populations, the populations that we serve. Um, but. Great workers want to want to be empowered and, and use the latest and greatest right? And so making sure that we are empowering them through being on the cutting edge when we can be and assessing those risks appropriately.
DAVID VENNERGRUND: Excellent. Great points. Alright. Uh, I, what I'd like to shift to a question from the audience. Do we have. Ability to, yeah, 
SORAYA CORREA: there comes surprisingly. Thank you. We do not have any questions from the audience. That could be, 'cause I may have lost connection on the iPad, but I know that, that some of our panels, 'cause we're running a little bit behind.
SORAYA CORREA: I know some of our panelists have to leave, so I want to thank you all. Exciting and interesting conversation. Um, I can't emphasize workforce enough. You, you know, that's near and dear to my heart. So thank you again for your insights, your invaluable insights, and I look forward to [00:33:00] the next conversation on this.
DAVID VENNERGRUND: Thank you.