Inside the Pentagon's AI Revolution

Bloomberg

Watch Inside the Pentagon's AI Revolution - Bloomberg on Dailymotion

Transcript

00:00This is the third story in our series on where artificial intelligence is already making a

00:05difference. Last week, it was teachers using AI in the classroom. This week is the effect it's

00:11having on the huge bureaucracy that is the U.S. military, where it's not so much what is already

00:16deployed as it is changing the entire theory of warfare and how to prepare for it.

00:22The future of war is going to come when you take that very large quantity of vehicles and robotic

00:28systems and marry it with an intelligence that can see, think, and act on the battlefield.

00:33It's really about a changing nature of warfare where we're looking at how to incorporate autonomy

00:38into all kinds of different operations. Warfare is going to be fought with a mixture of kind of a

00:44human and a machine. The U.S. military has long put a premium on avoiding mistakes at all costs,

00:51but with artificial intelligence, the government might need to take a page out of Mark Zuckerberg's

00:56playbook. Move fast and break things if it's going to keep up with technological change.

01:01We as an army have done an incredibly poor job over the last three or four decades of just saying,

01:08hey, if you have an idea that we think could be powerful for soldiers, get it to us as quickly

01:12as possible. The secretary of the U.S. Army, Dan Driscoll, is the point person for getting the

01:17Pentagon to take a whole new approach driven in large part by AI. The way that we used to acquire

01:23things as an army is, we'd have 16 steps that a thing would have to go through before we wrote

01:29a check. And any of the stops along those 16 could send it back to the beginning. And with the

01:34incentive structure where saying yes was punished and saying no was rewarded, most times it would end

01:40up in this doom loop of kind of forever decision making. And we are collapsing all of that down.

01:45So everyone will report directly to the chief of staff of the army and I,

01:49and we will hold them accountable for going very quickly and testing new things and learning.

01:56Former U.S. Department of Defense Deputy Secretary Kathleen Hicks agrees that these changes are

02:02essential, but she also warns that they're not easy. To what extent is there resistance in the

02:08Pentagon for really making the changes that AI may require? Culture change overall, I think,

02:13is really our biggest challenge. And it isn't just in the Pentagon. It's all across the stakeholders

02:20on Capitol Hill, throughout industry. There are a lot of invested incentives in doing things the way

02:27they've always been done. But AI is being used especially away from the battlefield in terms of

02:34bringing in lots of data and then using AI to quickly sift through that data and make sense of it.

02:40So if you think back, for example, to the wars in Iraq and Afghanistan where Americans face challenges

02:48around IEDs, these explosive devices that were often buried in the earth, you can imagine how AI

02:55is already being used to look at pictures visually to understand different data that's coming in.

03:02We really are just at the beginning of that maturation cycle where you could imagine different autonomous

03:09systems. I think that is the next frontier. Ryan Seng is the president and co-founder of one of the

03:17companies hoping to drive the change in the U.S. defense posture. Shield AI is an aerospace and

03:23technology company moving at breakneck speed to develop the AI-powered drones Secretary Driscoll says he needs.

03:30For the last 20 years, adversaries have modernized and enhanced their defense capabilities or their

03:37warfighting capabilities. And our ability to deter conflict in the future depends on the adoption of

03:44new technologies to make our warfighters more effective. And chief among them is AI and autonomy.

03:50What does AI make available that otherwise you would not have from other technology?

03:55I think the most fundamental thing that it does is it enables the deployment of effective mass on the

04:01battlefield. You can see in Ukraine millions upon millions of drones and missiles being produced,

04:09but they're limited in their ability to see, think and act based on what's going on in the battlefield.

04:14They might be remote controlled by a very focused operator who's connected to them via fiber optic cable.

04:19But this huge volume of robotic systems, whether they're drones, land vehicles,

04:23or boats or undersea vehicles, don't have their own ability to see, think and act on the battlefield.

04:28And then therefore their effectiveness is limited. The future of war is going to come when you take

04:34that very large quantity of vehicles and robotic systems and marry it with an intelligence that can

04:40see, think and act on the battlefield so that it is effective.

04:44If you think of having to defend against a swarm of 1000 incoming drones, a human brain is not capable

04:51of pulling off that decision making at that scale and the speed required. It's a really complex

04:57problem that just human beings are not well suited to answer on their own. And then if you think that

05:02you're in a wartime area and your enemy has those types of defensive capabilities that are run by

05:09artificial intelligence, it's going to be really hard for a human being to plan an attack in that space.

05:15And so in a lot of ways, what part of warfare may look like is artificial intelligence driven,

05:21drone on drone fighting may be the new future of the front line for a while.

05:27As Secretary Driscoll and his colleagues at the Pentagon spur the organization to develop

05:32high tech weaponry for the future, they're watching it get deployed right now in Ukraine.

05:38Ukraine is considered by many to be the Silicon Valley of war.

05:42We are hoping to repeat those lessons learned through our processes and our systems here.

05:46But what we do know is drone warfare is completely upending and altering how wars have been fought

05:53and how people have thought about fighting. We have got to get to a place where we can update

05:57things quickly. I was just a couple of weeks ago at a base and looking at one of our kind of air and

06:03missile defense systems and the laptop that was running this system was 30 plus years old.

06:09This soldier using it was 22. So this computer he is trying to use is eight years older than the

06:15soldier. You have to be able to update things within two weeks. And so it is not just a failed

06:20system. It is a sinfully failed system. There's been a lot of work from the U.S. military side with

06:28Ukrainians. Also, our NATO allies work closely with the Ukrainians. The Brits, for example,

06:34are very engaged in learning from what's happening there. The Russians are also learning and we have

06:41seen improvements from them. But I do think we're very engaged looking at what's happening in the

06:46Ukraine war and trying to learn our own lessons from it. It's not just AI and drones that are coming to

06:56warfare. It's also new technology like autonomous vehicles as German AV trucking company Fernride is

07:03demonstrating right now in Europe. Hendrik Kramer is the CEO. So right now we have this pressure cooker

07:11moment in Europe where the geopolitical situation and the war in Ukraine and the potential conflict of

07:17NATO in Europe with Russia is leading to a huge demand for unmanned systems. And ground autonomy,

07:24unlike the drone systems in the air, it has not been deployed and developed. Therefore, I think the

07:30impact will be broadly in defense and also civil logistics. So one of the most important defense

07:37application is very similar to a hub-to-hub autonomous trucking product, where you are, for example, having

07:44a coupling bridge between Poland and Lithuania, where Belarus and Russia are having this very small gap to

07:52connect the Baltic states with Poland and mainland NATO countries. And I think this is one of the

07:57applications where it will be very dangerous to put people into trucks on public roads. And therefore,

08:02this is a fantastic application where the same technology that is working for civil or defense

08:08or vice versa can be developed and scaled right now.

08:15It's one thing to see the future. It's another to move aggressively to reach it.

08:19And S.H.I.E.L.D. AI's Ryan Sang says there's still work to be done.

08:24If that is the future of war, how much of it is in the present? How much is AI already being used in combat

08:31situations? It's being used selectively today. It's not deployed at very large scale. And I think

08:37a lot of that is just the friction that exists between defense departments globally and in industry.

08:45If you look around the United States, I guess specifically, there's so many examples of industry

08:50moving out at light speed. And our own defense department has shown its capability to mobilize at

08:55light speed. But there has been a lot of friction in the acquisition system that slows down the

09:01government industry partnership. And I think that has been responsible for slowing down the adoption

09:06of A.I. despite many of the capabilities existing today and being battlefield ready today.

09:15As promising as A.I. is in giving the United States new warfighting capabilities, it is not a

09:21replacement for the soldier any more than it can be for your doctor or your teacher.

09:26We're going to need everyone. It's all hands on deck, as I used to say at DOD. We need our

09:31traditional defense manufacturers, particularly for the scale of manufacturing that we require for

09:38their knowledge and deep expertise. And we need that innovation that's coming all across the sector,

09:44but particularly from the startup community. At the end of the day, warfare has to remain a human

09:50active judgment. But A.I. can really help bring speed and precision to all kinds of aspects of military operations.

Category

Transcript

Be the first to comment

Recommended