AI and the Future of Aerospace & Defense
Insights
- AI adoption in defense is not a technology problem: organizations that fail to address workforce psychological safety and AI literacy before deploying tools will not realize returns, regardless of the quality of the technology.
- Studying how personnel actually use AI, which models they prefer, what questions they ask, and how long they engage, produces more actionable deployment intelligence than evaluating the models themselves in isolation.
- Waiting for a complete, clean data foundation before building AI is a primary reason defense programs stall; the more effective approach is to curate data for the specific use case at hand and build a shared foundation incrementally from there.
Recorded at the Mission Ready AI event in Melbourne, Florida, this keynote features Venice Goodwine, CIO and AI Product Owner at Arlo Solutions and former CIO of the U.S. Air Force and Space Force. Goodwine draws on 36 years of experience and direct responsibility for AI deployment across 183 installations and 800,000 personnel to argue that the primary barrier to AI at scale is not technology but organizational readiness: workforce psychology, AI literacy, and leadership alignment have to come first. She describes how she built NIPRGPT to study human-machine teaming behavior across multiple LLMs, established the Air Force’s first AI Center of Excellence to drive use-case prioritization and cost accountability, and used both to move from experimentation to production. Goodwine also addresses industry directly, arguing that partners should lead with capabilities rather than tools, arrive with pre-built coalitions rather than solo pitches, and use acquisition pathways like Trade Winds and OTAs to match the pace the mission demands. Drawing on her current work at Arlo Solutions, where she has applied the same methodology to a commercial AI-enabled cybersecurity product, she closes with a clear message: the strategy exists, the funding exists, and the guidance is public. The only variable left is the decision to move.
Venice Goodwine:
Today I want to have a conversation with you, and I want to converse with you about how I, as the CIO for the Department of the Air Force, drove our strategy for AI implementation. I’m going to say something very provocative right now in front of defense and aerospace industry. AI is not a technology issue. It’s an HR issue. If we don’t think about the psychological safety of our workforce in this era of AI, none of what we do is going to work. So when you go back to your organizations, please talk to your CHCO and say, “As I implement this new capability and people are worried about losing their jobs because I’m giving them some higher order thinking, here is how you comfort them.” So I just had that public service announcement out there, so when you hear my comments, you don’t gasp.
Venice Goodwine:
So let this slide sit with you for a moment. There is a sense of urgency that we need to think about because it’s real. When you hear us think about what I need to provide to the war fighters, the airmen, and the guardians as we deliver air capability, bombs on target, warheads on foreheads, whatever vernacular you’d like to use, we’re going to do that in an AI-enabled environment. So this is what the Secretary of War says about AI. It’s about overmatch. So I want you to know, yes, it’s great that our teams are using AI to create their meal plan, exercise plans. That’s great. But what’s more important is that our adversaries also have the capability, and it is important that we outsmart them.
Venice Goodwine:
It’s not an Air Force thing. It’s not a military thing. This is a whole of nation, so I need your help. So I’m also pleading with you today as industry. Let’s partner on what we call this whole of nation issue.
Venice Goodwine:
When you think about where we are, I’m not making political statements, but I will tell you 36 years in the uniform, and I will tell you as a CIO for two services, I will had to be wide-eyed about the technology that I was going to deliver. And so this is the type of sense of urgency that I speak of, the entire spectrum of AI. Yes, back office is important, but AI-enabled autonomy is also important. When I can send a drone or an autonomous aircraft and not send my airmen, my fighter pilots, I need to know that the AI works. And that’s what they’re asking us to do. That’s what they’re asking you to do today. So let’s talk about it.
Venice Goodwine:
Our war fighters need four things. It’s across our industry. We know what those four things are that they need. So I’m going to talk a little bit about what they are. Because whenever I sit and talk to airmen and guardians, they never say, “May I have access to ChatGPT? May I have Claude?” No, they’re not asking me that. They’re asking me, “When you deliver this capability and I push the button, can I walk away? Will it work? Can I trust it?”
Venice Goodwine:
So when you think about the capabilities that you are delivering and creating, because AI is in everything, right? Your lights, your refrigerator, your cars, you name it. But what I like is you have to have trust. And you know how we build trust, not only one with another, but how do you build trust with technology? How do you build trust with software? Because at the end of the day, AI is software. So how do you do that? So let’s talk about these four things that we need from a secure perspective, transparency, integrated interoperability, and then definitely let’s talk about trust again.
Venice Goodwine:
So when I say secure, secure by design. I don’t mean the ATO that you’re used to hearing about, “Oh my gosh, it takes so long to get an ATO. I got 353 controls.” Not that type of trust. Yes, that’s important, trust me. But what I’m talking about is the adversary has the same capabilities that we do. Are we going to outsmart them, out-design them, out-code them, out-innovate them, or are we going to wait and see what they do and be reactive and respond? Our war fighters that you're creating capabilities for, they want to know that we are first, and we have to give them that level of trust and that level of comfort in that.
Venice Goodwine:
But transparency is key. When I tell them that I’m AI-enabled, can I also tell them the data that it’s trained on? How many parameters I have? Can I tell them the true cost? Can I tell them rather there was drift, rather there was hallucinations? How transparent are you being when you talk about providing these AI-enabled capabilities? That is going to be important in our defense industry because it’s everywhere. It doesn’t matter whether I’m in a tank, whether I’m using night goggles, it does not matter. AI is there to help. So I need to make sure that the soldier or the guardian that’s managing our constellations, that they truly understand that I can trust the output that comes from what I’ve given them, because I’m changing the way we operate and how we do business. So that’s important.
Venice Goodwine:
But it has to be integrated. Remember, we do joint fires. You’ve heard the term CJADC2, yes? Combined Joint All-Domain Command and Control. You know why it’s combined? Because I fight with my nation and my partners, my allies. So why it’s joint? Because it’s all of the services together. Why is it all domain? Because it’s land, it’s sea, it’s space, it’s all. And command and control gives us the ability to make decisions faster. AI enables that, but it has to be interoperable. I cannot operate in a silo anymore. Yes, I’m Air Force and Space Force former CIO, but I will tell you, the Army CIO, the Navy CIO, the Coast Guard, we all conspired and colluded together to make sure that our capabilities, while different for a ship and for a F-35, even though both services have them, but still, we talked about interoperability because that was important.
Venice Goodwine:
Then it goes back to trust. When I use AI in a command center, in a command-and-control operations center, in a decision support system, and a response comes back that I should do B instead of A and C, can my operators trust it? Can the Pacific Command commander in Hawaii trust it? Because he has an adversary that’s right there close to him. So when I tell Admiral Paparo that, “Yes, you can use this AI-enabled decision support tool." He wants to say, "Got it." And so that’s what we have to think about. These four things. Every decision, four words, it’s very important.
Venice Goodwine:
So what gave us permission? What gives you permission to move? Every time a senior leader speaks, it is communication to you of what we’re thinking. It’s communication to you of what our problems are, what we’re trying to solve. It’s not because we like to be on stage. It’s because we are trying to communicate with industry the hard problems that we have. Because we tend to look at things through the lens that we’ve always looked at things. But the beauty of what you do, you have so many verticals within your companies. You have so many customers that you interact with. Imagine the ingenuity that you have, that you bring to us. So when you get that opportunity to sit in front of a senior leader in any of our services, tell them the problem that they need to solve that they don’t even know they have. That was beautiful to me as a CIO. I wanted to hear, I wanted to listen. I wanted to hear you say, “Here is the problem that you have, and here’s why the solution that you have is no longer working, and here’s how I can integrate or complement what you have.”
Venice Goodwine:
Power to move. You have power to move. You have the power to move. Because when I have conversations, strategy drives investments for me. But for you, strategy drives R&D investment. So I communicate to you on a level to tell you what’s next and to tell you what’s coming, so that you could spend your R&D dollars in the right way.
Venice Goodwine:
So think about the DoD AI strategy that came out. Everyone in the room, raise your hand if you’ve read it, yes? Please say yes. Raise your hand if you didn’t, just so make me feel good. Makes my slide work when you participate. But yes, you read it, because they tell you about all the pace setting projects that they’re working on. Think about it. They just gave you the answer to the test. I’ll talk about that a little bit more.
Venice Goodwine:
The Air Force just issued a data AI strategy. Started when I was there. It’s now complete. Why? Why does data and AI go together? But here’s the problem I will tell you, and I want you to tell your industry partners, and I want you to tell the senior administrators when you go talk to them. Don’t try to curate all of your data now, and at one time. I know, blaspheme. But that’s what we’re trying to do. No, I want you to curate the data for the use case in which you’re building, because guess what? Then when that data’s curated, the next use case may use that data. It’s done, and you build upon it. The problem, the reason we’re frozen is because we’re saying, “My data’s not ready. Oh, my data is unstructured, it’s semi-structured.” It doesn’t matter. Choose the use case and the data you need for the use case, and then curate that data. Trust me, this is why the data and AI strategy goes together.
Venice Goodwine:
But I think there’s some lessons to learn from EU. I love that their structure of assigning risk levels to the models and how it’s going to affect the citizens in the EU. So flip that and think about when you provide an AI model to me or any of my counterparts, label it in a way that, how is it going to affect the war fighter, the asset? Just take it all the way back and think about it in that way. So I think we can learn a lot from that. No AI strategy without data. But it doesn’t have to be all the data.
Venice Goodwine:
I want you to be clear that when you come to the Air Force, Space Force, the Army, and you talk about your AI problem, you don’t have to solve it end-to-end. Be specific in the problem you’re trying to solve.
Venice Goodwine:
But I’m going to tell you a little trick here. Come closer. Shh. No one’s going to tell you this. I’m going to tell you because I’m not there anymore. It is going to be important that when you come and talk to me or anyone in my position, that you already have established partnerships. That’s not new to you in this room, because a lot of you are partnered together. But here’s what I mean. This administration, like others, do not want vendor lock. So when you come to me and you tell me that, “Well, Ms. Goodwine, I can do everything for you.” Can you imagine hiring the plumber who’s also going to be your electrician, who’s also your architect? You would never do that, ever. So don’t sell to me that way. So what I ask you is when you come and have a conversation, and you have a problem to solve, you may be a part of that problem. Bring your partners with you. Already established a partnership. Work in the back room of who’s going to get what share of the profit. But when you come together holistically with already established partnerships, and then you can talk about the problem end-to-end, you’re winning. Those that talked to me and did that are the ones that got me to do the North and South. But when you come to me alone and say you can fix everything for me, you got the tilt of my head. Like, “Hmm, really?” So partnerships is very important when you think about the use case.
Venice Goodwine:
You say, “How do you know?” Because I did it. Here’s my receipts. I did it for the Air Force. All 183 installations worldwide, 800,000 CAC holders, 12,000 cyber and IT professionals, and I had a $17 billion budget. I was successful because the first thing I did was, I understood AI literacy. Because there are people that you’re going to put this tool in their hands, and they don’t understand the tool. So it starts with foundational.
Venice Goodwine:
But let me tell you, I did this. This is a success because I made a mistake. How many of you heard of Reddit? Yeah. Let me tell you what Reddit is for the Air Force and the Space Force and the Army. It’s where the military, the uniform people go when they don’t like something that’s happened, they can say it anonymously. They put it on Reddit.
Venice Goodwine:
So OpenAI, ChatGPT came out. I didn’t understand it. So what did I do? I wrote a memo and blocked it. “Thou shalt not use ChatGPT on airforce.mail.” Oh my goodness. Reddit let me have it. “I can’t believe her that she would do that.” And I said, "Oh." I had to clutch my pearls. I said, "Well, let me go back and get smart about this." Because that was not the right decision. Because I didn’t ask the very people that I was going to put it in their hands. I didn’t understand their literacy. So I went back to the drawing board and said, “Well, let me educate you on this.” Just like before I give you a car, I make sure that you train and I license you. So I did that with AI. Wow.
Venice Goodwine:
You know what happens when you do that? The ingenuity that comes from that, the innovation. Fail fast. Try it, didn’t work, doesn’t matter.
Venice Goodwine:
So then I created NIPRGPT, along with Alexis Bonnell if you’ve heard of her. She was at Google, and she’s at OpenAI. I give her that credit because the one thing we wanted to understand with NIPRGPT is the human and machine teaming. It wasn’t about just the tool itself. It was about if I created a platform that allowed different models to be changed in and out, I could see how my airmen and guardians behaved. I could see the questions they were asking of the model. I can see the results. I can see who choose which model, who had preferences. So we switched out. Yes, we used ChatGPT. Yes, we did use Claude at the time. We used all of the models, and we changed them out. I was interested in token cost, I was interested in the questions they were asking, I was interested in the responses they were getting, and I was interested in how much time they spent. NIPRGPT is now what you call GenAI.mil. That is what the Department of War is now using to do the exact same thing. More than a million users today on Gemini. They started with first the Gemini, the Google platform. And they too are adding models for the same reason. But it started with NIPRGPT because we needed to understand what our users actually needed.
Venice Goodwine:
But let me tell you why I established the Center of Excellence. Because when you’re an organization this large with a budget this big, you cannot waste money. Because one thing about being a public entity is there are some people on this thing called the Hill, that they want to know every penny you spend. And so I had to create a Center of Excellence, but it was designed to do a couple of things. One, yes, I had training, but it served more as an open door for use cases. Yes, use case come in, requirement comes in. I decide the use case. We write it up, which model works best, because remember, we did NIPRGPT. I understood investment. I understood who would use it. All that in one center of excellence. It made sure that we had the best return on investment for our AI dollars. But it also helped me work with you better in this Center of Excellence. So if you talk to the data and AI officers today in the Air Force, they still have that.
Venice Goodwine:
Now, why do most efforts fail? You heard me say fail fast. Do not let perfect be the enemy of good. If you do that, you will not deliver a product, because guess what? Every time you do a pilot and you put a bunch of users in there and tell them to work on your tool, they’re going to find something wrong. You’re going to create a backlog. You’re going to keep trying to fix it. No. Give an MVP and move on. Fix it in the backlog. Don’t let perfect be the enemy of good. And the department understands that.
Venice Goodwine:
This is why they have programs like Trade Winds. Everybody familiar? Prototype. Trade Winds is the best way to go. One of the best ways. Why? Because in a five-minute video, the capability, someone likes the capability, someone with budget authority, acquisition authority, and own the requirements, got to have the triad, can now use your prototype. So take advantage of Trade Winds. SBIRs. I love the new law that just came out. You can go to a phase three. If I can get a customer to sign a memo, I can go to a phase three SBIR. That’s with appropriated dollars, as opposed to small business dollars for phase one and two. Use these things to your advantage. OTAs are now our friend. All of our contracting people in the world. OTAs are now your friend. So it allows you to do the prototyping.
Venice Goodwine:
So again, don’t wait for perfect. Don’t try to sell a tool, sell a capability. One of the major things in the Air Force we do, it’s called category management, and everything I do, I put it in categories. If I buy a software, it has to line up to a capability because I have to be able to rationalize that category. You know what I mean. This is endpoint security. Every tool for endpoint security is in the same portfolio, so to speak, so I can manage investments. You know the Air Force, the military, we buy things in force, right? Your PACE model - Primary, Alternate, Contingency, Emergency. Once I get four things in a capability, if I want the fifth thing, I need to start doing rationalization, because why do I need five endpoint protection agents? Just an example to tell you that sell us capabilities. Understand, if I’m telling you I’m going to do joint fires, if I’m telling you that I’m working on the kill chain, then what capability helps with that? Don’t lead with the tool. Lead with the capability. I think that’s important.
Venice Goodwine:
And then don’t treat security and speed as opposites. Secure by design. It was an initiative long time ago, but if you’re using DevSecOps, the CI/CD pipeline, you’re doing this. And you know what happens when you do this? When you’re done, you’re already in an accredited stack or credited cloud. You’re in the commercial cloud, the gov cloud, your FedRAMP is easier. Your ATO is done because you inherited all those other things from your IaaS and PaaS. Do once, use many. So don’t let security be the opposite. It will help you. That’s why most fail. But again, it worked for me, tried and true.
Venice Goodwine:
I left the Air Force a year ago. I work now for Arlo Solutions, a small, a medium business. And what am I doing there? I just created an AI-enabled security tool, cybersecurity tool. And everything that I told you I did, I did at Arlo. It’s called DecisionBeacon, cutting through the cyber noise. Why? Because first I educated AI literacy, data and AI strategy. All the things that I told you worked in the Air Force, they work where I sit today. Solutions, cafes. I love it when I read on the Infosys website. I love to hear the conversation prior to this on what you’re doing in that space, having a harness, an orchestrator, being able to trust, transparency in your models. All of that is very important.
Venice Goodwine:
And what made the difference? Just a pictorial view for you. This is what made the difference. I started before I was ready. After I pulled down that memo, by the way. I just want to let you know, that did end up in a good news story. I got rid of the memo. I rescinded it, said, “Good men, I’m sorry.” But this is what I did. I said, “Go. Here’s your tools. Everyone that had a common access card that can access the network, go play.” I even allowed others who are not Air Force and Space Force because I needed that much data to make a decision. People first, people mission, common tools, set the pace, be first. That will make the difference.
Venice Goodwine:
But here’s what I’m going to tell you about you. Here are the DoW pace setting. If you have not read the strategy, I put this here just for you because this is exactly what A&S, I’m sorry, our acquisition and sustainment. Here’s OSW, Pete Hegseth. This is what he cares about. He wants AI, and AI-first at wartime speed. And they’re in these categories. Do you see yourself in any of those categories? AI-enabled everything. Yes, productivity is important, but I need the warfighters to be able to make decisions faster, execute mission faster, and I know that AI is a tool to help us do that.
Venice Goodwine:
Now, it is not the panacea. It is not for everything. We must be very careful in how we use it. But here is your authority to move fast, came straight from the Commander-in-Chief to the Secretary of War, to all military departments and field agencies. Guardrails. No constraints or restraints, but also responsible AI. We will use AI ethically. And the power at the tactical edge. Where’s the tactical edge? Wherever my warfighter is. If he’s wearing night vision goggles, that’s the edge. If he’s operating in some austere environment, that’s the edge. If he’s sitting at his desk on an installation, that’s the edge. The edge is where the capability is, and the warfighters need to have the confidence that they have in the tool that you provided to them.
Venice Goodwine:
And I’ve already talked about GenAI. But integration and interoperability is key. I don’t want to glaze over that. That’s important because we don’t fight alone. Because not only with the mil deps, and now we start to bring in our NGOs. When you think of where we are today, it’s not just one service. It’s not just uniformed personnel. Interoperability is important. Our countries, and allies and partners are at different technological maturity than we are. So understanding the interoperability and integration is important.
Venice Goodwine:
So here’s our next move. Here’s your next move, but we’re going to do it together. Because again, cannot do it without you. You’re the defense industrial base. I cannot do it without you. Rather, it’s maintenance on the asset or enabling the asset. Here’s your next move. The only thing we need to do now is move. We have a strategy. We have guidance. We have money. We have willing participants. We have a mission, and we have a sense of urgency. The next thing we need to do is on us, and us means me and you.
Venice Goodwine:
So I hope today that I’ve at least shared with you from my perspective, from the Air Force perspective, the space perspective, and I will tell you, if you lift and shift that to any other service, you will hear the same thing. And I think if you lift and shift that and take that to your organization, it will be same and true for you as well. So thank you today Infosys for giving me the opportunity to share and speak with you, and thank you all for playing along with me today.