This is Neil Earwoods at the Center for New American Security. We're talking today with Greg Allen, who's a new adjunct fellow in our Tech and National Security program. Thanks for being here, Greg. Great to be here.
And he's the author of a new extensive report, Artificial Intelligence and National Security. Let me start with a kind of obvious one. We hear a lot about how AI is going to reshape the economy, the workforce, day-to-day life, but we hear a lot less about how it's going to affect national security. So how is it going to reshape, as you argue, national security?
In a lot of different ways. And we, in the report, try to profile it across three different dimensions. The first is actual military superiority, which is basically can we make better weapons, can we defend against those weapons more effectively, called by economic superiority, because ultimately the root supporter of national power is usually the economy. You're only as powerful as your economy can support.
Exactly. And then, finally, information superiority, which deals with two aspects, both our ability to enhance our intelligence and espionage capabilities, but also the pace of innovation itself. And whether or not a lead in this technology makes you more likely to discover leads in other technologies that then reinforce, once again, military superiority and economic superiority. So you argue in the report that AI is going to be as transformative as nuclear weapons, along with a long list of other big changes, and that's obviously a big claim.
What makes you think that this is going to be that big a shift? Sure. And we are unabashed in the report about going forward and saying that. Specifically, we say it's going to be as big a change as nuclear weapons, aircraft, cyber technologies, and bioweapons.
And we looked at those technologies specifically because we were interested in changes in technology that were revolutionary, not merely different. And basically, everyone that we talked to said that AI was on that same level, such that the fundamental paradigms and approaches to national security that nations used might not hold in the same way. So an example of the type of things that we think are revolutionary across AI is the extent to which it might change the underpinnings of power and technology. So we are used to technologies in the United States that are both expensive and complicated, meaning that the only people who are likely to have them are likely to be great nation states.
And even then, only great nation states were really willing to spend what it takes to have the advantages. So an example is stealth. It cost billions of dollars to develop. It was an advantage that lasted three decades.
And if we had it and you did it, the odds are we're going to win. And billions were in overruns alone. Exactly, exactly. And so those are the types of technological advances that we are used to.
What's different about artificial intelligence and the reason why it's going to be revolutionary is that while there are dimensions in which it makes great powers a lot stronger, there are also dimensions in which it makes great powers a lot more vulnerable. Because as prices fall, and as most of the most interesting technological advances take place in commercial industry, the nature of the relationship between the public and private sector from a national security perspective changes dramatically. Well, and I wanted to ask, particularly since, as you said, the vast majority of the leaps forward are coming in the private sector, how are we going to be able to incorporate those advances in the private sector, particularly since they're not coming from companies that are always working with the federal government, how are we going to incorporate those advances into our traditional national security? Yeah, I think this is an important starting point.
I think a lot of people still have a fantasy that maybe there's some super-secret laboratory somewhere buried in the NSA that is actually a really advanced AI research laboratory and way ahead of commercial industry. That is simply not the case here. But I like that fact. I know.
It was really reassuring in some sense, maybe worrying in some other senses. But the fact is, it's simply not true. The vast majority of the advances in basic AI research and development, so sort of fundamental leaps in technological approaches that are relevant to a lot of different interactions and applications of AI, those advances almost exclusively are being made by the private sector, which is not to say that there's not great work going on in academia that's sponsored by the government, but there really is a lot more work being done in the private sector than we are used to here. The DOD is paying attention to this.
They are not ignorant of this fact. And their official response is that they're going to develop an edge in applications of artificial intelligence, which is to say we will not necessarily be in the government side of the leaders in the basic AI R&D, but we will, among nation states and among people interested in military affairs, be leaders in instantiations and applications of AI for national security purposes. I think the problem with that approach, and I don't think it's necessarily wrong, but the problem with it is that so much of what is being developed in the commercial industry is a relatively easy port over to national security. Some stuff is quite hard, but some stuff is quite easy.
If long-distance drone package delivery becomes a reality, we're talking over tens or hundreds of miles, then what you have also facilitated is that anybody who has the capacity to buy or make an IED and provides explosive device a drone, and suddenly they have a poor man's version of precision strike weapons. And so I think the difference in artificial intelligence is that the technology is so advanced, so capable, but also, given the cost trajectory, is likely to be so cheap and so accessible. And that is what is so unfamiliar to the United States, to have leading-edge technological capabilities that are definitely relevant to a warfighting context be available in the commercial market with minor modifications. Well, what you're talking about gives me tremendous anxiety.
So how does the U.S., how should DOD prepare for this kind of leaps and bounds change? Sure. And this report, I should mention, was commissioned by IARPA, the Intelligence Advanced Research Projects Activity. It's the sister agency of the military, IARPA is the intelligence community.
But IARPA really encouraged us to take a whole-of-government frame. And so the recommendations that we have in the report are very much looking at not just what the DOD can do, but maybe what the State Department should do, what relative to the relevant scientific R&D centers in the government should do. I think our first overarching recommendation is that there's a paradigm shift, which is we need to say, once again, as I said before, artificial intelligence technology is not merely different. It is revolutionary.
And so you need to understand the technology as something where a lot of your old assumptions are going to need to be revisited, a lot of your old paradigms are not going to work. The costly and complicated versus cheap and simple is one of those types of paradigm shifts, but there are many others. And so when you adopt that different paradigm and you look at the existing portfolio that the DOD has across this entire military asset base and technological asset base, some of the stuff upon which we rely very heavily today might not be relevant. If the long-distance precision strike of a drone becomes cheap and just take commercial hardware and lightly repurpose it, if it gets distances in the hundreds of miles, then stuff like aircraft carriers, which we're used to worrying about for ship-bored ballistic missiles, might actually be vulnerable for much similar attacks, might be vulnerable to terrorist organizations or other smaller activities.
Or a guy in a fishing boat. Exactly. And so I was talking to a friend of mine last night, Matt Harris, who's very sharp on these issues. He was giving me the example of Billy Mitchell, who was a person in the Army Air Corps in the interwar period.
He was one of the champions of air power, and he needed to buy the budget. And so what he did is he had an experiment where he took his plane out and bombed a ship, a decommissioned German World War I ship, and proved that airplanes could sink battleships and said, air power is what matters. Give me the money that you are currently giving to the Army Coastal Defense. And so while I'm not arguing for any kind of budget robbing, I am arguing for that type of paradigm, which is look at what the technological trajectory suggests.
Look at what capabilities that we know are going to be possible. And then think about how that's going to reach a military power from a perspective of both what do we need to have in that world, but also what don't we need to have in that world. What do we currently spend a lot of money on that might not actually be that useful? So that's one thing, is that we need to be having these thought exercises.
And when we think about what type of speaking is needed, our specific point of reference is the RAND Corporation during World War II, which is you have this dedicated entity with experts across a lot of different disciplines who are thinking about this night and day, because there are so many paradigms that are so baked into these military organizations that really do need to be revisited. Well, so, Greg, thanks so much, Greg. Hey, I appreciate it. Thanks.