Deputy Defense Secretary Kathleen H. Hicks and Jim Mitre Conduct a Fireside Chat at the Virtual Department of Defense Digital and AI SymposiumContracts for June 8, 2022Defense[...]

JIM MITRE: Good afternoon. My name is Jim Mitre, and until recently, I was a senior advisor to the deputy secretary of defense. My job was to help establish the CDAO, and the CDAO, the chief digital and artificial intelligence officer, just hit full operating capability last week.

Now, in the Department of Defense often times after a mission, there'll be held what's often referred to as a hot wash, which is an opportunity to reflect on the mission, tease out some of the key lessons learned and think about the implications for the future.

So in the spirit of a hot wash, the deputy secretary and I are now going to engage in a fireside chat to review some of the thinking and really understand her observations and insights about the CDAO.

No further ado, it is my distinct privilege and honor to introduce the deputy secretary of defense, Dr. Kathleen Hicks.

DEPUTY SECRETARY OF DEFENSE KATHLEEN HICKS: Hi. Good afternoon, Jim. And I note there's no fire, but I'm letting it go. We'll fireside chat regardless.

(LAUGHTER)

MR. MITRE: Indeed. Great. Well, one of the things that I'm interested in understanding a little bit more, and I expect many on the line are as well, is some of the initial observations you had when you first came into the Department of Defense and were thinking that there is a specific problem here that you were trying to solve with the CDAO. How do you think about articulating that problem?

DR. HICKS: Sure. So as you know, Jim, I'm at my heart sort of a strategist, and I think in a systems-level approach. So I certainly didn't walk into the department and say, "I'm a software program or a data analyst," or "I know how to build -- you know, I'm a software engineer." It really was coming in, seeing all the things we want to achieve for the nation, for the nation's defense, and then understanding the kinds of concepts and capabilities that lead to that, and how -- what opportunity we had here, based on what had been built before us in data, in the digital space, and of course, on A.I., and how we had this opportunity to really move forward on our mission space if we could capitalize on those moves that had come before us. So that was sort of the underlying sense I had.

And the more I talked to the communities of interest, learned what they were doing, got excited about the types of capabilities they can bring to bear and tie that, if you will, bottom up. I don't mean that in any disparaging way, but from that sense that they had at that operational level of what they could provide with those big, strategic problems that we were looking at. It was a very natural next step.

We took that step with a lot of advisement. We talked a lot with Congress. We talked with folks on the outside, the -- certainly, the A.I. Commission folks. We -- I asked for -- commissioned two separate independent studies that what commercial best practice was, and certainly, talked to our own workforce. And coming out of that, really, it became clear that something like what we now call CDAO was the best way to think about the tech stack that would help us unlock a lot of what we need in the department.

MR. MITRE: Excellent. One thing that you've talked a lot about is the importance of using data-informed decision-making to provide you and the secretary with decision advantage, if you will, from the board room to the battlefield. I'd love to understand a little bit more about how you think about using data and some of the observations in terms of trying to work with where the department is today on having data inform your decision-making.

DR. HICKS: Absolutely. So Jim, you know all too well, I'm an analyst at heart, and I like decisions to be based on facts and knowledge, and we should all hope that that's the case. We sit -- I sit, as the COO of DOD, on top of the largest organization in the world, and we are under-gunned in terms of the analytic capability that we tend to bring toward problems relative to the scale of both our size, and then the kinds of the consequences of the challenges we're looking at.

And so data is another way -- an avenue -- toward better analysis, better fact-finding, understanding how to see ourselves, understanding how to see our adversaries and all the other facets of a situation.

So if I want to ask a question on sustainment, where 70 percent of our funds in acquisition aren't going toward a platform; they're going to the sustainment side, I need to have an understanding of where that money goes and what the drivers are, and how we can fix it.

If I'm asking a question, or the secretary is asking a question about the logistics flow of materiel into Ukraine, well, let's look at the data of where that logistics flow is. What are our munition stocks? What are the risks?

Data informs all of that, and I think what we've been able -- because we are stepping into leadership in this period of time, we're able to show that promise, you know, the little peek under the tent about what unlocking data can do. And then if you can add predictive analysis to it, what if we could really analyze all that data in a way that's speedy with A.I., and also even then, be predictive?

That shows so much potential, as you said, from the board room inside, as that sustainment example shows. And then, as you get closer in the, you know, sensor-to-shooter piece of it, you start to see the advantage for data and for A.I. and the CDAO construct, again, as that tech stack comes tighter together and there's a virtuous feedback cycle in really giving the United States that advantage on the decision-making side, accuracy-prediction side.

MR. MITRE: Excellent. Now, as a strategist and analyst, it's not lost enough that while you were thinking about creating the CDAO, you were also working on the National Defense Strategy. I would love to learn a little bit more about what you see as the key elements of the strategic context that should inform the CDAO as it embarks on its work.

DR. HICKS: Absolutely. I -- as I hope my examples I just gave demonstrate, almost any strategy the United States would have right now would benefit from a CDAO, so it's a -- I think it's an enduring concept, and again, builds on what's come before, and I would anticipate in the future anyone would continue to build on this, no matter what their NDS is.

That said, this NDS is very much aligned to what we're trying to achieve with the CDAO standup, and of course what the broader community is trying to achieve across all the digital and data spaces.

So that includes, for instance, in this strategy, the three ways that we go about the strategy. The first and most overarching and integrative is integrated deterrence. And there, we're focused on how do we bring combat credibility - how do we bring combat power - across the spectrum of conflict from below the sub-conventional level up through the strategic level across all the domains; air, land, sea, space, cyber.

How do we work with Allies? Well guess what, data is -- how we share data, how we operate together with a common operating picture, as we might have called it a mere 20-years ago, that is vital to integrated deterrence. And I think we, as I said in the sensor to shooter side, there's a lot opportunity here that enhances U.S. combat credibility.

On campaigning, which is our second way in the strategy, that's really about day-to-day how do we dynamically shift the agile - make decisions, be opportunistic and frame that deterrence, the integrated deterrence in a day-to-day way that tells an adversary today is not your day. You don't want to go after the United States today.

Doing that really does again get enhanced if you have more decision making time and space, better indications and warning, faint signals, being able to read faint signals from a broad array of different sectors and data; economic, informational, certainly military. So it's really inherently-important there.

Finally, building enduring advantage, our third way in the strategy is about making sure that we make the investments today, the seed-corn today is invested in so that it harvest - we can harvest it in the long term. That's about our work for - the space we're talking about, it's about making sure we have a strong workforce, we're technically competent, we have the processes from hiring processes to procurement processes in place to be agile.

We know how to work with the commercial sector well, and this is a space where I think, again, we want to demonstrate we can do all of those things. Both with our own workforce, our own contracting but also as part of a broader U.S. economic competitive space.

MR. MITRE: Excellent. Well, tying to themes together, you talk about the strategy, you made reference to the National Security Commission on A.I. report; that was in many ways a touchstone for us in developing the CDAO. Regularly reviewing, consulting its recommendations. And you talked about the important of thinking about A.I. with the tech stack, right. And bringing CDAO is in many ways brining some of the key building blocks together of that tech stack.

The NSCAI commission report also talks about the importance and the urgency of the department's pursuit of A.I. technologies in relation to the China competition. Is there anything in specific that's coming to your mind when you think about the NSCAI recommendation or the China dynamic here that is important for us to understand for CDAO?

DR. HICKS: Well let me just say two things that might not immediately come to mind for folks. But the first is, what it told me is we need world-class A.I. talent. And that explains Mr. Craig Martell as a selection for CDAO and the great team that he has around him with Margery Palmieri and others.

But that's what it told me right there which is the United States and the Department of Defense specifically needs to bring that kind of talent. We need to be able to access that talent, not just hire it.

But we do need folks inside the system able to help us conceptualize and build out the kind of , you know, A.I. enterprise that we need.

I think the second thing I would go to, again, probably not necessarily where others might immediately is the importance of the U.S. using A.I., employing, having A.I.-enabled approaches that fit with our norms and the way we think about warfare. And that's where responsible A.I. is so important. We have right now underway a rewrite of the DoD policy on responsible A.I.

We have been a leader in responsible A.I. We have set the pace for Allies and partners and certainly for adversaries who have nothing like it. And I think we can be very effective in warfare. We can create an effective deterrent and stick by our norms. And that's who we are as a people. That's really important to me that we advance that through CDAO.

MR. MITRE: Absolutely. And in talking with you you're starting to scratch out a theory of change management - of how you move the bureaucracy. I'd love to understand a little bit more about what your theory is. How do you - given the extensive experience you have here -- how does CDAO, Mr. Craig Martell, come in actually move bureaucracy in a meaningful way?

DR. HICKS: I think leadership is incredibly important. And the ways in which leadership speak, use their time, drive the system, hold people accountable, will be very important here. The secretary and I are ready to be at the forefront of that, to be part of that change, and to help Mr. Martell do the same along with the team in CDAO.

So right at the top is leading, and how we set our expectations and hold folks accountable. Part of holding folks accountable is eased substantially in terms of cultural change by aligning the incentives correctly.

That can be very, very tricky here in the department because we are - we have many cross-cutting incentive structures. We have within Title 10, different subcomponents of DoD that have different mandates, as opposed to a corporation in other words, where it kind of cleanly comes up to the top.

We of course have, across the river, we have folks who have different views potentially and different from one another, about what they think the department should be seeking to achieve and have the ability to, you know, line in and line out what our direction is. And so that creates some complications.

And there are others that are complicated but I think that's the heart of how you change culture, is you go after the incentives. So what's that theory for CDAO so far? And I am definitely going to be looking to Dr. Martell to help us with his insights on how he thinks about change.

So far, part of our theory is showing particularly combatant commanders or commanders at the operational level what they wish they didn't - couldn't live without. What can they not live without? And, you know, often times they have no idea. They're perfectly happy with a grease board, just like I - people might have been perfectly happy with a printed map to get them through cross streets in Washington D.C., you know, 10years ago. Whereas today you can't live without app and in particular if your app is crowd-sourced and it has lots of information and things like that that help you navigate.

Similarly what we're finding is if you can show those use cases and they get back to what I said before, they get to the heart of the mission. And our mission is both to obviously deliver to the war fighter, and it's also to deliver to the taxpayer. So we're back to your opening - one of your opening question. We want to go after both of those because the incentives inside our structure are different for different communities.

So if we can save a lot of money, which we will be able to save a lot money. If we can pay back through investments in our tech talent and our tools and our upgrades of software, you name it, we want to go after that, to get some of that money back, because that money can help the warfighter and it also goes to our mission as a responsible steward of taxpayer dollars.

But more than anything what we want to do, is be able to find those use cases and unlock the potential for decision makers like the secretary out to the field to the warfighter, to show them how they can achieve their objectives - their military objectives, operational objectives on behalf of the United States - better, smarter, faster with with this toolkit.

MR. MITRE: OK. Probing just a little bit further on this topic, you mentioned our friends across the river. One of them is the Congress. And one of the things CDAO is going to need to do is try to find a way to help establish constituencies that can help get more attention on things like software as opposed to hardware and move in that direction.

What advice do you have for Dr. Martell and Margie and the team in terms of how to build those types of constituencies that can actually meaningfully change the direction the nation's going in?

DR. HICKS: Sure. Well, there is a massive U.S. innovation ecosystem and it is increasingly focused on software rather than hardware, and that's true, of course, for defense as well.

And so the ability to point to what the department can ignite in that industry and how we interact with industry, small businesses all the way up through, you know, large global businesses, I think that's a really strong place to start, demonstrating that we are part of an American competitiveness story, that it - we don't have to think about it as the defense industrial base, if you will, per se, although they're a big piece of how we go after this issue set, but we have to be really thinking about what DOD can do on behalf of the American competitiveness story. So I think there's a lot of opportunity there.

I think second, I would say more functionally, you know, it's talking, it's telling the story, it's explaining those use cases I pointed to, that this isn't a science experiment, that we can demonstrate value, value can be counted in dollars, and value can be counted in mission success and lives saved. And we will absolutely - we are already - have some of those examples, from Afghanistan and Ukraine just within the last year, to work. That's been undergo - underway, for example, under the JAIC, with GAMECHANGER and some of the other efforts that are longer-standing. We just need to be better at telling that story.

MR. MITRE: Let's talk a little bit more about industry. One of the things we heard a lot in establishing CDAO, from small businesses, you know, large, traditional primes, as well as the venture capitalist community, is a real desire based on mission to try to work with the Department of Defense, but yet still a lot of challenges in being able to access things, given how long it can take for contracts to be awarded, given the opaque nature in certain cases about the regulations processes they need to abide by.

How do you think CDAO should play in that space and help establish stronger relationships with industry?

DR. HICKS: Let me first say I feel in no way defensive on this question and no one in the department should. We are extraordinarily difficult to work with. And there is a lot for us to shift in terms of our processes, some statutory issues, but mostly in our processes and culture and workforce training and staffing. I mean, a - particularly post-COVID, you know, we have some real staffing challenges throughout the department that then have compounding impacts on this issue of how we work with industry.

So I think the first thing is, as in any 12-step program, acknowledge the problem. So we need to acknowledge that we are a difficult partner and we need to get better. Then, I think it's about how we go after it.

I think there's a real - part of the impetus I had in CDAO is there is a power in bringing a vanguard organization with direct reporting relationship to me and to the secretary at the four-star level that can push us in these areas.

Again, they can build on work that's been underway. We have a number of, you know, procurement vehicles, for instance, already available. I think there's five that are really focused on expanding the - our access in DOD to non-traditional companies.

We have the ability to be a vanguard on the workforce - the innovation workforce - and I have an effort underway on that. I'm very hopeful that CDAO will be right up there with ideas of how we get better at both recruiting talent, but upskilling and reskilling and then retaining some of that talent. Some talent, we need to be really comfortable with it flowing in and out. I think that model is completely appropriate, and we just have to - we're not comfortable with that typically, so we need to get more comfortable with it.

So those are some of the ways I think CDAO will be an exemplar. And again, I'll just say vanguard, cause it's more than an exemplar. This ought to be one of the places the folks look at and say "that's the way we want to do this in government, this is government going the right way" and it's - hopefully we'll be able to do it at a scale that then translates - helps us translate across the whole department.

MR. MITRE: OK, excellent. Yesterday in one of the panels, we actually talked a bit about some of the obstacles CDAO could face in making the department help be better at adopting DOD analytics and A.I. at scale.

You initiated at this forum a year ago the ADA initiative, which is all about accelerating A.I. and data...

DR. HICKS: Advantage.

MR. MITRE: ... advantage. Thank you.

DR. HICKS: No worries.

(Laughter.)

Yeah. The good news is I remember. It would be bad if I can't remember what ADA stands for, yeah.

MR. MITRE: With that initiative - you know, it's been a year since you've announced it and there's been a lot of efforts underway in that regard. Are there some key lessons-learned, in terms of what's working or where you see more opportunities for CDAO to help advance ADA and deliver analytics and A.I. at the commands?

DR. HICKS: Yeah, I'm very pleased with where we are a year in. That said, you know, we can do more and we will do more, and we're always chasing behind the curve, I think, because of the pacing challenge that we face with China.

On ADA, what we found is there's a lot of enthusiasm out at the commands when we can get the teams out there. COVID, again, has been complicating to that, easier more recently. And we see a lot of natural-use cases, questions that combatant commanders really want help answering, and the ability to apply answers through ADA, which, by the way, is named after Ada Lovelace, I have to stress, because I've heard other people refer to it as "A-D-A." So I wish - we wish to express right here it's named after Ada Lovelace.

Another thing we've seen is that there is - unsurprisingly, there is different maturity levels throughout the department, of course, but also, of course, throughout the combatant commands. And so some are - had already been further along their own journey on adoption of both A.I. and data than others. And so we're kind of - we're, by necessity, sort of pacing to that.

And then last, I'd say we're finding that many of the problems that we're seeing are common and/or there are common solution approaches. And because we're at the enterprise-level at ADA, that's kind of the beauty of the federated approach. We have this centralized repository of knowledge and expertise and data and tools and contract vehicles and folks who understand how to use contract vehicles for this purpose, and then these problem sets can come in and we can tailor, if you will, more easily and get solutions out faster. I think that's what we've seen to date.

It's a three-year effort, meaning I'm funding it, if you will, for three years, and we're going to see aswe get through F.Y. '24, being the last of those years, where we are with ADA and what the next national evolution is.

I think we should be very unafraid to shift approaches as we stand-up of the CDAO itself shows, and make sure we are ahead of the curve, not chasing a curve by being committed to either particular initiatives and/or to organizational constructs.

That said, again, I think ADA in and of itself is proving its worth, and I think anything that would follow it, ADA - whether it's called ADA or something else, naturally will build on what we do here.

MR. MITRE: Great. Another one of your key initiatives is related to responsible A.I.

DR. HICKS: Yes.

MR. MITRE: Put out ethical principals. You, through the CDAO construct, have actually elevated the role of responsible A.I. within the organization by having a chief of A.I. assurance, who is a direct report to the CDAO, and will soon being putting forward an implementation plan on our A.I. What are your thoughts on how that's going, how the department is actually moving towards taking that seriously, and being more responsible in the development of A.I?

DR. HICKS: I do think we're a leader in responsible A.I. The department got out early, and then again under Secretary Esper put out some guidance, and again I put out some guidance last year, and that has the benefit of demonstrating both from the very beginning, if you will, back in 2012 when we last worked together, we were putting out responsible A.I. guidance and that we are continuing that commitment. And that that commitment is bipartisan, that it's happened under both the last administration and this one. That's the good news. I do think we're leading.

I think the areas to grow are the international community probably needs to push along further in this space. We will want to be a leader in that, but we also want the overall space of A.I. use to be governed responsibly. And there is a lot of room there. I think again with CDAO being established at the four-star level really elevating that responsible A.I. initiative and bringing it into focus, I hope that there's a lot of opportunity there to grow.

MR. MITRE: OK, good. Now with CDAO just achieving a full operating capability in many ways now the hard work begins for the organization. How are you going to evaluate its progress, and what do you have in mind when you think about what success looks like for CDAO?

DR. HICKS: Sure. Well I do think it's really important to empower the team there to, first of all, take a little time as they put themselves together in FOC, to think through that problem set themselves and put some measures out that we can talk about. So I'm not going to get in front of that process.

What I will say, is that we have to be able to deliver - we have to advance, and advance quickly, on the challenge set that the warfighter faces. And so, inside the department often this is all characterized under JADC-2, Joint All Domain Command and Control. However one wishes to characterize it, what we want to be able to do is make sure we are leveraging the state of the art in order to increase accuracy, increase speed of decision making, increase the quality of our ability to deliver effects. And if we can do that, it helps, again, all the way back up to our mission. It helps us defend the nation in particular by demonstrating our ability to do integrated deterrence, enabling integrated deterrence, and enabling campaigning.

We really can't campaign effectively. We have tried - globally - against a competitor with such significant work underway in the military and in other realms as China - just sort of hand-jamming and PowerPoint-sliding our way through that. We need to be able to access the data that's out there, but we need to do it our way, not China's way - back to the responsible piece.

And so, we need to be able to have a system and approach within the department, and the CDAO has to be seen a year in as delivering on that as the go-to place for talent and technical expertise to get after that problem.

MR. MITRE: Excellent. OK, well unfortunately that's all the time that we'll have for today to explore this topic, but thanks so much for finding the time to chat with me, and share a little bit more of the logic behind the CDAO.

DR. HICKS: And thank you, Jim, for your hard work standing up the CDAO. We would not be here today if it were not for all that you did, so thank you.

MR. MITRE: That's kind of you to say. Thank you.

Public link

Please use the above public link if you want to share this noodl on another website.