Transcript: Robert Work on “Intelligence Matters”
In this episode of Intelligence Matters, guest host Admiral James “Sandy” Winnefeld (ret.) speaks with Robert Work, the 32nd United States Deputy Secretary of Defense for both the Obama and Trump administration. Work and Winnefeld discuss the Pentagon’s “Third Offset” Strategy, and delve into the military applications and ethical dimensions of technologies like artificial intelligence and quantum science. They also review the Defense Department’s transition from focusing on counterterrorism and counterinsurgency to great power competition. Work, now the Distinguished Senior Fellow for Defense and National Security at the Center for a New American Security, explains how Russia and China are developing a range of technologies in an effort to leapfrog the U.S. in the military realm.
Military applications of new technologies: “We don’t know how AI and 5G and quantum and synthetic biology, we don’t know how they are all going to go to work. But they all have the capabilities to provide a step function in the way we fight wars. And the competitor who gets there first is going to have an enormous advantage. So if you’re a force designer, you’re happy. Because you’re buying new capabilities. This is a time of enormous foment inside the department. And it’s a pretty exciting time.”
The stakes associated with AI: “[T]he competition in AI is a central one in great power competition between China and Russia. AI will reflect the values of the competitors. Whereas we want to protect human privacy, we want to protect human dignity, we want to make sure that our use of AI is ethical and moral and consistent with our laws, an authoritarian regime might not do it that way.”
On competition with Russia and China: “This is not a time where we can really afford to waste the time we have. We believe that the Chinese and the Russians are really pressing us in the military sphere. They’ve had 18 years of kind of coming after us while we’ve been focused on counterterrorism. And so they’ve closed the gap to an uncomfortable degree. And so what we have to do is really move out quick.”
INTERVIEW WITH ROBERT WORK
CORRESPONDENT: SANDY WINNEFELD
PRODUCER: OLIVIA GAZIS, JAMIE BENSON
Welcome to Intelligence Matters with Michael Morell. I’m retired admiral Sandy Winnefeld, filling in for Michael while he’s on travel. Our guest today is the honorable Bob Work, who served as the 32nd deputy secretary of defense from 2014 to 2017. Secretary Work served as an artillery officer in the U.S. Marine Corps, and worked at the Center for Strategic and Budgetary Assessments, and as CEO of the Center for a New American Security.
He also served as undersecretary of the Navy, and is chairman of the board of the U.S. Naval Institute, among many other activities. My close association with, and deep admiration for Bob began when he was deputy secretary, and I was vice chairman of the Joint Chiefs of Staff. These two positions work very closely together, principally driving the formulation of the department’s budget. Bob, welcome to Intelligence Matters. It’s so great to have you here today.
It’s great to be here, Sandy. It’s great to see you again.
Bob, one of the things I quickly came to admire about you was your immense grasp of the history of the department, and of war fighting writ large, beginning with your description early in our relationship of the Third Offset, which drove much of our thinking in 2014 and 2015. Can you walk us through the progression of those offsets?
Sure. Well, since World War II, the United States has really counted on technological superiority over its adversaries. It’s been baked into its strategy. It’s been baked into its operational concepts. It’s been baked into its tactics. And after World War II, when the United States came face to face with the Soviet Union, it realized very quickly that the Soviet Union outnumbered us in tanks, in airplanes, in soldiers.
And there was just no way for us to match them, tank for tank, plane for plane, or soldier for soldier. So we sought to offset their numerical conventional superiority in other ways. The first offset took advantage of our then nuclear superiority, and said the way we will do this is to develop battlefield atomic weapons. Cannons that would shoot atomic warheads, rockets, missiles.
And this was very, very successful for about 20 years. The Soviets were deterred in large part because they knew that if they did try to attack western Europe, that we would use nuclear weapons early. Now, by 1973, the Soviet Union had gained strategic parity with us. They had as many nuclear weapons as we did.
And the threat to employ atomic weapons early in a war just didn’t make any sense anymore. So once again, to offset their numerical advantage, we went after what I’ll call battle networks, which are these large digital systems of systems that have a censor grid that looks deep behind enemy lines, a command and control, an intelligence grid that makes sense of the information coming in, and writes orders to units and decides how they want to apply effects on the battlefield.
And effects grid that employs effectors like missiles, bombs, electronic warfare, cyber attacks. And then a regeneration and sustainment grid that keeps everything going. But the key to it is everything we were shooting were guided weapons, which actually hit what you shot it.
In unguided weapons warfare, the dirty little secret was almost everything you shot missed. And as the range went up, the miss distances went up. So collateral damage was just the order of the day. Guided munitions would actively correct their trajectory so that they would home in on the target, and hit very, very close to it.
So it was the combination of these battle networks and the guided munitions that in this vernacular of the day, allowed us to look deep, shoot deep and kill deep. And once again, that deterred the Soviet Union conventionally. Now, when I came aboard as the deputy secretary, and you were the vice chairman, all of us were very concerned on the trajectory of the military technical competition.
Both Russia and China were rapidly catching up with us in guided munitions and battle network warfare. So the third offset was materially different than the first two. The first two were trying to offset just numerical advantage. The third offset was saying, “What happens when the bad guys can look deep, shoot deep, and kill deep as well as we can?”
And we quickly concluded that we needed to change the way we thought about fighting. So the Third Offset was all about that. And after a lot of discussion and analysis, we decided that the way we would approach this is to start to inject AI-enabled autonomous capabilities into our systems, across the board. And we figured that that would allow us to regain a measure of military technical superiority.
Fascinating. And brings back many old memories, I might add.
So how have your views, Bob, evolved since those early days in 2014, 2015 when we worked together? Have events confirmed your views on the progression of how our approach to technology should evolve in the Third Offset?
I think so, yes. I’m kind of a glass half full type of guy. So I see a lot of the things that are going on in the Department right now, there’s an awful lot of experimentation on robotics, on autonomous systems. So in that regard, I’m very gratified that we seem to be making some movement.
But I can flip this around, and be a glass half empty, and say, “Look, the Third Offset was announced back in 2014. Here we are six years later. And I’m really not satisfied on how far we’ve come.” There’s an awful lot of competing technological advancements out there, 5G for example, which we will use in our command and control systems.
Quantum science, which will allow us to do better machine learning and better communications and better censors, synthetic biology. We’re in a technological tsunami right now. And going back to being a glass half full type of guy, it’s important for us to try to figure out how they all fit together. And the competitor who can fit them all together in a way that provides competitive advantage are going to be the ones who can really set the pace in the competition.
It would seem like there are probably some considerable service cultural obstacles, departmental cultural obstacles, congressional obstacles that are lined up, making it harder and harder to do exactly what you’re talking about. Maybe we’ll get into that in a future segment of this particular episode.
Bob, we mentioned a moment ago artificial intelligence as part of a Third Offset. I’d like to turn now to your considerable expertise in artificial intelligence, including your service as vice chair of the bipartisan National Security Commission on Artificial Intelligence. Can you baseline our audience on what artificial intelligence really is, how it breaks down, and a bit about the commission’s work?
Artificial intelligence has gone through what have been called AI winters, where the promise of AI, where it would allow us to do so many things, have robots that would speak with us, and do what we would want, systems that would automatically take care of everything. We’ve just never had the capability.
The way DARPA, the Defense Advanced Research Project Agency, describes this is that the first wave of AI were what were called expert systems, if/then logic. And it was quite powerful. But you had to have, like, a Ph.D. in computer science. You had to have someone who really understood the problem and could program the system.
And you could only go so far. But along comes machine learning, where you can actually teach the machine, or the machine can learn. And the second wave of machine learning, which really took off around 2012 and 2013 is really what is happening inside the department.
So you can provide the machine with a lot of information. And the machine can make inferences, and actually make predictions of the future. They’re prediction machines. So you can use it for readiness. You can say, “Look at what is happening in my aircraft fleet. And I want to improve readiness.”
And artificial intelligence can learn through just going through enormous amounts of data, and saying, “This is the proper way for you to approach this problem so that your readiness would improve.” You can use it in the back office of the Department of Defense to improve transactions.
Anything that you can think of in terms of transactions, you can do. Alan Turing, who is an expert in this, says, “If it takes you about a second to think about something, you should just automate that.” Third wave is where you’ll get into contextual AI, where the machine can actually understand human intent, and try to figure out what is the best way to do a problem.
So the one thing, though, is in the Department of Defense, AI is just a means to an end. The end is more autonomous operations. And the department spoke in terms of autonomy at rest. These would be decision making tools, or decision assistance tools, or tools that would allow an analyst to understand enormous amounts of data that he or she might not otherwise be able to really get it through.
And then autonomy in motion, which would be unmanned systems, autonomous weapons, et cetera. So it was the idea of having all sorts of autonomy, which would allow your battle network to operate at a much higher speed than legacy battle networks.
And that is where we thought we would gain our advantage. It made sense for a U.S. battle network because we value initiative, as you know. We expect the youngest soldier, if they’re on the battlefield, and see something that looks different than what the orders originally envisioned, they’ll take the actions necessary to get us back on track. So by the combination of human initiative and more autonomy, we thought that we would really be able to have an advantage over battle networks designed by authoritarian regimes, that don’t really appreciate initiative.
It’s interesting. Now Bob, one of the key questions surrounding the use of AI in defense systems is, of course, ethics.
People are worried that we’re headed down the road of the infamous Skynet from Terminator movies. The department has held the line on having humans in the loop for any lethal firing decision. But that might have to change out of necessity. How do you look at the ethical dimensions of AI? How do you even define that as a problem?
Well, I should have probably started by saying the competition in AI is a central one in great power competition between China and Russia. AI will reflect the values of the competitors. Whereas we want to protect human privacy, we want to protect human dignity, we want to make sure that our use of AI is ethical and moral and consistent with our laws, an authoritarian regime might not do it that way.
You can see the way that China’s using AI-enabled systems to provide surveillance over their population, and actually control them. And in the western part of China, the way they’re using it on the Uyghurs, a Muslim population in western China is really quite concerning.
So from the very beginning, we have emphasized that artificial intelligence should be based on some ethical foundations. And these ethical foundations are being discussed in a wide variety of different fora. And it’s one of the key things that the National Security Commission on AI has said. Because this is a competition in values as much as it is a competition in technology, we want to make sure that our ethical foundation is a strong one.
It would be interesting to see if there were ever a point when the advantages conferred on an opponent by using AI start to so eclipse our own capability that we have to consider looking at other ways. Speaking of China, Chinese entrepreneur Kai-Fu Lee says that to have good commercial AI, you have to have four things, a lot of government support, decent engineers, ruthless entrepreneurs and lots of data. Given that, and given China’s somewhat serious advantages in commercial AI, can we compete with them in the defense space in AI?
Yes, absolutely. Now, Kai-Fu Lee also says, look, he thinks that the United States and the West is ahead in AI research, but that he assesses that the Chinese are ahead in applications, for the very reasons that you just outlined. And the one thing that I would say is I don’t think we are being ruthless enough in the applications.
Everyone is very worried that AI is brittle. Once you teach a machine something, it can’t transfer that learning very easily. There is a technique called transfer learning that allows you to do this. The best example I can give is DARPA used to show a picture of a young baby holding a toothbrush. And even after being taught, the machine identified it as someone carrying a rifle.
And no human would ever make that mistake. But the machine did, because it was just trained, and it made its best inference and guess. So what we need to do is start making applications, testing them, gaining trust in them, experimenting with them, finding the limits of what these technologies will do, and making sure that we apply AI in a reasonable way, so that what we expect the machine to do will actually be observed by our operators.
It actually would be against international humanitarian law to give an autonomous weapon, for example, to a commander, and say, “Utilize this weapon.” And the weapon would do something different than what the commander was thinking. It might wind up causing collateral damage, or killing or maiming an innocent non-combatant.
So it’s very, very important that we get into the applications phase, and we just push and push and push. And we find out the limits that we’re comfortable with that are consistent with international humanitarian law, and the law of war. And we use it to the best advantage.
I’ll give you an example. The U.S. Army just recently did what is called a breach of a complex obstacle. So this would be an obstacle that an infantry unit might face, or an infantry-mechanized unit might face. And you’d have anti-tank ditches and mine fields, concertina wire.
Just a very, very difficult tactical problem. And we train our troops to, over a period of weeks, how you would have to do that. And the Army recently did a completely robotic breach, everything. The breach was done completely with robots. And normally this takes a long time to train.
But after it was all over, the senior leaders of the Army went down and said, “Hey, how long did it take you to train?” They said, “Oh, not long, sir.” It’s because the Army used Xbox controllers as the basis for the controls for the robots. And all of the soldiers immediately just got it. They got the Xbox controller and went to work. So that’s the kind of thing that we should really be doing at all levels, and trying to figure out how humans and machines work best together to accomplish a complex battlefield problem.
Bob, no contemporary discussion of defense can escape the transition we’re going through in shifting our focus from counterinsurgency and counterterrorism to near-peer threats. Can you reflect for our audience a bit on what drove this and where it’s headed?
Yes. The Cold War ended, and the United States found itself in an enviable position. With its direct treaty allies, allies that were obligated to come to our defense if we’re attacked, and we would be obligated to come to the assistance of our allies if they were attacked. When you added that together, between our percentage of global gross domestic product, and global defense expenditures, both were over 70%.
It was a period of enormous advantage. The United States faced no competitors that were really capable of knocking it off of its position as the number one global superpower. So throughout the ’90s, and especially after the horrific attacks on 9/11, the United States really focused on two things, rogue nations such as North Korea, Iran, Iraq in the 1990s, and then after 2001, terrorism.
And it really focused on these things to the exclusion of really large powers that could take them on conventionally, take us on conventionally. Sandy, since 1886, the United States has never faced a competitor with a gross domestic product greater than 40% of our own.
In the Cold War, the Soviets got to about 43%, 44%. Even in World War II, the combination of Japan and Germany was not that much more than 40%. China has surpassed us in purchasing power parity, and is expected to surpass us in absolute GDP in the 2020s.
So we’ve never faced a high tech authoritarian competitor with the economic power of China. Russia is a different problem. They have a terrible economy. But they have a lot of nuclear weapons that threaten the destruction of the United States. So over time, both China and Russia said, “Look, we cannot allow the United States to be unfettered in the world. We want to compete directly with them.”
And so starting in the late ’90s, both of the countries said, “We want to compete directly with the United States. And we want the challenge them on the global stage. We don’t like what the United States is doing. And we want to make the world safe for authoritarianism.” If the U.S. fought the Cold War to make the world safe for democracy, both China and Russia are trying to undermine democratic ideals, and raise up authoritarian regimes.
And so around 2012 was the first time the department said, “We need to start paying attention to these great power competitors.” It started with the creation of a very small organization called the Strategic Capabilities Office, which was set up by Ashton Carter, who was then the undersecretary of defense for acquisition technology and logistics.
And he said, “You know, I don’t like what I see in the western Pacific in terms of military technical trend lines. I want the Strategic Capabilities Office to address that.” Then in 2013, there was a very famous National Security Council briefing, again, given by Ash Carter.
Now he’s the deputy secretary of defense. And it was on the threats to our space constellation posed by both China and Russia. And from that point on, we started to develop capabilities that would protect our assets in space and threaten our potential competitors in space.
And this started with the Third Offset, where we said we have to look at these great power competitors. We have to do something different. And I think you can draw a straight line from the Third Offset to the 2018 National Defense Strategy, which was written when Secretary Jim Mattis was the secretary, which basically said, “We are now going to focus on long-term strategic competition with revisionist powers like China and Russia.”
So this is kind of like a bear coming out of hibernation, and shaking its head. It’s been asleep for a long time, didn’t really need to have strategic muscles, because it was so big there wasn’t really a competitor that could threaten it.
So now we’re shaking off our kind of 25 years of having it easy. And we’re trying to build our muscles back up so that we can compete in the long-term strategic competition. And the United States, Sandy, as you know, is pretty good at this. We did it pretty well in the Cold War.
And we ended the Cold War on terms favorable to us. So this has been a major strategic shift. And it doesn’t come fast. As you know, turning the ship of the Department of Defense takes a little while, and takes a lot of work. But I think everybody should recognize that work is happening. And there is a major, major shift afoot.
So this discussion leads me to ask a question that’s very important to me. I usually think in terms of balancing ends, ways, means and the strategic or the security environment. And if one of those variables changes significantly, the others have to adapt in order to maintain equilibrium.
Well, if we would agree that the security environment is deteriorating a bit, and our national ends would remain the same, protection of our allies, and certainly our own homeland, that would leave only two variables to tinker with if the security environment deteriorates, which would be ways and means.
And as a military, as a nation, we tend to focus a lot on means, better technology, more of it. And we tend to ignore ways. Do you see getting to a point where the Chinese become so competitive, both in closing the gap in conventional capabilities, but also coming up with new asymmetric capabilities, that we have to turn away from long-standing operational concepts, and move into completely new ways of thinking about this competition, certainly on the military side, but using all instruments of power?
Yes, that’s happening right now, Sandy. Secretary of Defense Mark Esper has challenged the joint staff and the J7 specifically, saying, “By December of this year, I want you to provide me with new operational concepts that would help drive the design of our force, the capabilities we pursue, and the way that we would execute military operations, along the full range of military operations, from competition in the gray zone that a lot of people talk about, all the way up to conventional warfare.”
The department recognizes that the way we have done business in the last 25 years simply won’t be able to be replicated in the next 25 years. Because again, both China and Russia said, “Look, the United States has this enormous advantage in guided munitions and battle network warfare. So we’re going to make our own battle networks. And we’re going to be able to have technical parity with them in this type of warfare. And if we can, we’re going to try to leapfrog them.”
Going back to our conversation on AI, the Chinese believe that artificial intelligence is the way they leapfrog the United States in the military realm, as well as the economic realm. So everyone recognizes we have to change the means. And this is requiring a lot of work. There is a lot of experimentation going on in the western pacific with things like what are called multi-domain task forces, task forces that might see an Army unit using anti-ship missiles to target ships at sea, all sorts of different things.
So that shift is occurring now. It’s an exciting time, both strategically, if you’re a strategist you’re excited because grand strategy is cool again. If you’re somebody who likes military technical revolutions, you’re happy because we’re on probably the cusp of another one.
We don’t know how AI and 5G and quantum and synthetic biology, we don’t know how they are all going to go to work. But they all have the capabilities to provide a step function in the way we fight wars. And the competitor who gets there first is going to have an enormous advantage. So if you’re a force designer, you’re happy. Because you’re buying new capabilities. This is a time of enormous foment inside the department. And it’s a pretty exciting time.
And, of course, speed to that capability will be a very, very important function of this, with a very, very active competitor.
Bob, I’d like to start this segment by talking about space, which is, of course, a very intense ground for competition between the United States and other countries around the world. The realization that our space command and control capability was both weak and vulnerable was an unforgettable moment for me a few years ago. And you really seized on this. Can you walk us through how space has evolved for the department, particularly while you were in office, but even since?
Yes. Let me go back a little bit further. And this was in the Cold War, when we thought there was a missile gap. But we couldn’t get deep into Soviet territory to really get any intelligence to figure it out. We developed high flying spy planes, like the U-2. But ultimately they were shot down.
And we went into space. And throughout the Cold War, space capabilities were referred to as national technical means. Because they were there to support the national command authority, to provide indications and warning in case we were under impending attack, and also to carry us up to and through a nuclear exchange.
So it was very, very focused on strategic intelligence. When the Soviet Union went away, and in Desert Storm, we turned all of the capabilities of the space constellation in support of General Stormin’ Norman Schwarzkopf. And it was a big wakeup call.
Everyone said, “My gosh, the capabilities of our space constellation are unbelievable.” So we went through this phase where we started to use space capabilities in support of our joint war fighters, command and control, precision navigation and timing, weather, intelligence, surveillance, reconnaissance.
And it was an enormous advantage for a military that has to go across the Atlantic and the Pacific to get to a fight. They established their battle networks in the home field advantage of another competitor. So it was a sanctuary. There were no competitors who could go after these capabilities.
And we got lazy. And meanwhile, the Russians and the Chinese said, “This is an advantage we cannot allow the United States to have.” So they started developing a wide range of capabilities that threatened all of our space capabilities in all orbits, geostationary orbits, at 22,000 miles, high altitude orbits, medium altitude orbits like the global positioning system satellites, and low earth orbits where all of our intelligence surveillance and reconnaissance, or a great deal of them are.
And it really became clear that our constellation was under threat. And we really hadn’t paid a lot of attention to defending it. And we had better get busy with it. So when you and I were in the department, and soon after deputy secretary Carter gave his pitch to the National Security Council, we started investing in a wide variety of capabilities that would allow us to maintain our space capabilities even when they’re under attack.
So one of the things that you and I worked on was a thing at the time. And this is a mouthful, the Joint Interagency and Combined Space Operations Center, or JICSpOC. Interagency means that all of the intelligence community assets and DOD, Department of Defense assets would be worked together and combined, were our allies.
And essentially what it did is it looked up. And it would say, “Okay, we’re under attack. What are we going to do? We’re going to have to move this space asset so that it doesn’t get taken out, et cetera.” And over time, that has become the National Space Defense Center, which is in operation today out in Colorado Springs, Colorado.
So yes, this is one of the things I’m most proud of, one of the things that you and I started, and I think have set us into a position where our space assets are more protected. I would like to say this too. All of our war games say that if a war ever did start, and I’d like to emphasize a war between the United States and China, and the United States and Russia, is not foreordained.
We think about deterring the war all the time. None of us want to fight a war, I think. But all of the war games show if a war starts, the war will generally start in operational domains where attribution is difficult. Space, undersea, going after undersea cables, for example, and also cyber.
And those are the three areas where you really have to pay attention to. Because you might find yourself under attack. And you’re not really even realizing it until the competitor or the adversary really has gained an advantage over you. So this is a very, very important aspect of the competition.
It will be very interesting to get your views on the Space Force.
Bob, I wanted to give you a chance to reflect on the future of the department. It looks like budgets are going to be a bit flat for the foreseeable future. But the threats aren’t going away. The department’s had some reorganization. But plenty of the more senior positions are held by acting officials. There are a lot of vested interests among the services, COCOMs and Congress. What kind of institutional changes do you see necessary for the department, given its size and institutional complexity? And can they be accomplished?
I’m probably a radical in this regard. I believe really radical change has to occur. And it’s a longer conversation on what some of those changes might be. But the key thing right now is we have to be bold. You talked about speed. This is not a time where we can really afford to waste the time we have.
We believe that the Chinese and the Russians are really pressing us in the military sphere. They’ve had 18 years of kind of coming after us while we’ve been focused on counterterrorism. And so they’ve closed the gap to an uncomfortable degree. And so what we have to do is really move out quick.
So I would hope that the department really goes after bold changes, and not let the bureaucratic inertia of the department hold it back. I also think that Congress can help a lot. We’ve had too much reorganization in the department. Reorganizations which have caused, in my view, a delay in us getting after the things that are important. So I would say this is the time for bold action. And I hope to see it.
Bob, thanks so much for spending time with our listeners today. It’s been a very informative discussion. It always is whenever I’m with you. And once again, thank you so much.
Thank you Sandy. It’s always a great time to sit with you and talk over these issues.
* * *END OF TRANSCRIPT* * *