For our next session of the morning, we are honored to welcome the Honorable Emil Michael, Under Secretary of War for Research and Engineering and Chief Technology Officer for the Department of War, with David Sanger, Chief National Security Correspondent for The New York Times, to discuss the war-fighting technologies of the future.
Well, I'm back, but with a different inquisitor here. So, I am extremely glad that you have joined us here. And for those of you who don't know Emil Michael, and I think that's almost nobody here, he is the Under Secretary for Research and Engineering. And I guess the first question I want to ask you is, you've now, over your career, I think when we first met you were still at Uber doing that. You've come into a lot of different places, you've evaluated where their technology was, you've decided on big leaps they needed to take. Tell us how this one's been different.
What's been different is a few things. So, we've learned a lot from the Ukraine-Russia war about the power of drones and counter-drone technology and the speed of innovation required for both sides to keep up on advancements in the battlefield in real time. We've also learned a lot about the difference between an exquisite system that we might have and the world moving toward a more mass, attritable, low-cost, easier to produce set of systems, which is sort of a derivative of the Ukraine war, but also you see new companies like Anduril and these companies coming to the fore with a new way of thinking about a clean sheet weapons design. So those two things, and then we were behind in some areas, whether it was shipbuilding, hypersonics, these things I think the last administration completely ignored to a large degree and so we're kind of doing a lot of cleanup on that. And so really with the president's approach, with disruptive approach I'd call it, and the secretary's approach, we've been able to do a lot more faster in this moment of change, and including AI in that too, with AI where AI is now sort of it's entered a public consciousness. There's billions of users on AI within consumer and enterprise use cases and those three things, the factors have conflated into we're in a moment of great change and opportunity.
Great. We'll go through each one in the little time we have, but let's start with drones. So in the Biden administration there was a project called Replicator which was an effort to, you know, try to go toward less expensive, you know, drones like this that were supposed to be delivered, I think last summer was my recollection. By the time that deadline came along you had already sort of moved beyond that into much bigger programs and involving swarms and so forth. So tell us what the evolution has been from one to the other and if in fact there is a step function between those two.
Well, the difference is we had to do two things. One, we established a drone dominance program which is a focused program on we've got to do drones in a way that it delivers us mass. And then we've got to do it with our own supply chains or allied...
Mass, deliver mass of destructive capability or mass of numbers of drones?
Mass numbers. Right. And we've got to do it in our own supply chains. And the supply chains for drones, whether it's Chinese batteries or and so on, were highly distributed around the world. So, the focus has been both to have mass numbers and to have self-reliance within the US and within our allied partners. So, we created drone dominance and there was a lot of rules and regulations with something called the blue list where units can only use drones off a certain small number of listed drones which didn't allow the innovative startups who had new products to come and get on that blue list very easily. So, we made it more distributed for people to try and test a wider set of systems that were validated. And so, what we're doing now is doing sort of a contest-like approach to drone dominance. And the value there is that we have to have a number of companies that get enough orders to succeed because the innovation cycle is so fast.
Right. And if it's distributed in the way it was, which was very thinly across...
Then no one gets the demand signal that can invest in factories, that can invest in re-domesticating the supply chain. And so, we don't want to create one winner, we want many winners, but it can't be infinite winners because they'll all lose because there's not enough demand to buy. So, that was the structure of what we've done in drone dominance is we've had these things called gauntlets where we do tests and the 12 companies that make it out get some funding and then we're going to have it, you know, go down to five, then probably five is about the right number.
Probably about the right number. Okay.
And then they can innovate on a faster cycle after that and they could afford the facilitization that all requires so that we could produce them here at home.
So, you often heard the Pentagon across administrations, even back in President Trump's first administration, talk about the need to keep a person in the loop. And yet the more I read and hear and study what you're doing with the swarms, when you're thinking about discriminating on a target, it's hard to imagine that any individual could even keep up with it. I mean you could be on the loop but as something mostly as an observer. So is your concept that a human has to be behind each targeting decision or simply that a human has to be behind the design of the parameters that that swarm is going to pursue?
Well, there's a lot of different ways to interpret human on the loop. We have a directive called DOD 3009 that's been in place for 5 years which talks about what are the rules of engagement with humans in the loop, on or on the loop, on different kinds of operations and it largely depends on the operation. I'll give you an example which is if we're defending against a drone swarm and you have a directed energy laser or high-powered microwave that can take them down and do a better job than a single human, well, people would say, well, that's a good use of autonomy there, right?
Absolutely, and there's no human death in that so it doesn't even really qualify.
There's certainly, you know, fragments and things that can drop, but yeah, it's more precise. And I used to have this debate in the Uber days about self-driving cars. Is a self-driving car safer than a human? Well, we knew an Uber was safer than drinking and driving your own car.
And I think over time you'll see self-driving cars prove that they're safer than human-driven cars. And I think you could use that example for drone swarms. You could also use it for Golden Dome. Golden Dome for America is about protecting the homeland against missile threats. And some of the missile threats that have evolved over the last several years, just years, hypersonic missiles that are maneuverable, that can go five to 10 times the speed of sound, so that the amount of human reaction time that you have against something like that, well, maybe you could put something in space that can take them down. And Golden Dome's looking at that kind of technology. So, those kinds of things are lower stakes than when humans are in harm's way on the front end. And so, you'll have different sort of echelons of where you want humans to interact. But so far, besides defensive systems where it's clear that a human can't defend against a swarm or a hypersonic missile in the time required with human sort of making the decisions.
So, how about on the offensive side? Could we reach a point to your mind? And the example you were giving with self-driving cars, I think was a really good one. And it's great that you've, you know, bringing to it the Uber experience here. Could we reach a point where not only would an offensive autonomous only weapon be more accurate, but it might be more ethical?
Of course. I mean, I think there are those times. And that's why I say that there are different operations, you know, and this is the point. As warfare evolves, what you want to use in terms of the degree of autonomy changes.
