00:00The U.S. Navy is turning to artificial intelligence to speed up that process,
00:03awarding Domino Data Lab a contract worth nearly $100 million to help unmanned underwater vehicles
00:09potentially change that timeline from months to weeks.
00:12Joining us now is Thomas Robinson, the Chief Operating Officer of Domino Data Labs here with us in New York.
00:17Great to have you here on set.
00:18Thank you for having me.
00:19Let's start with that.
00:20What is your company able to do?
00:21We've heard from former admirals about the perils of these mines and the difficulty of removing them.
00:27We've heard about how there were demining vessels that have been kind of offloaded by the military.
00:31So this is a crucial and difficult problem at this point.
00:33What is your company able to do to expedite things?
00:36Right.
00:37So we're a software company, and we build a platform and AI solutions.
00:41And the particular thing that we're helping with in the Navy is updating their ability to use AI models on
00:50unmanned underwater vehicles.
00:52Everybody's seen the movie Titanic.
00:53You can imagine.
00:55Underwater vehicle, right?
00:56Those vehicles have sonar, vision, other sensors, and that comes in as a stream as they pursue the waters and
01:04look for objects.
01:06You know, historically, that was reviewed by sailors, right?
01:08By humans.
01:09By humans, right.
01:10They would scrub through and see, you know, all the particular things and identify.
01:14You know, maybe they had a sheet to identify what a particular mine or weapon looked like.
01:19That's a great task for AI, right?
01:21The AI allows you to rapidly speed up.
01:24You can consume and review all of that in one short period.
01:28And, you know, the AI can be trained to make fewer mistakes.
01:31Humans get tired, you know, eight, ten hours on the job.
01:34Maybe, you know, you're making it.
01:35It's monotonous work.
01:36Right.
01:36Exactly.
01:37So, you know, that's essentially what the AI is doing.
01:40We were also talking to one of our colleagues who was in the strait and was explaining how, I mean,
01:44we've been covering this, but I had assumed it was a more complicated operation than putting a mine on a
01:50small speedboat and literally dumping it off the side, which I would imagine makes it really hard to track where
01:56they're doing that.
01:56Can you do a surface-level analysis as well, or does this help make up for the fact that it
02:01is so hard to track where they're putting these, especially in a strait that we understand is very hazy and
02:06visibility is very low?
02:06I saw your reaction to that yesterday.
02:10You know, I noted it.
02:12Yeah, I mean, that is the challenge, right?
02:15This is a massive asymmetry because you have mines that are not very expensive.
02:19And then, yes, putting them off the side of a speedboat.
02:21But on top of that, the adversaries are constantly updating their techniques.
02:26Maybe a mine goes inside a 55-gallon drum.
02:29You know, what does that look like on the seafloor?
02:31How do you know that's a weapon or not?
02:33So, you know, that's where the technology is so powerful, especially this ability to refresh rapidly to adapt to adversarial
02:40and different theater changes, if you will.
02:43We talk a lot about, I guess, the big names in AI that get a lot of attention outside the
02:47context of military use, just generally speaking.
02:50And there's always been this speculation, when are we going to move to the tier beneath that?
02:53Other companies like yours that are doing this kind of important work.
02:56I wonder if you could just shed some light on your capacity to pivot so much as you have to
03:01work on national security issues.
03:03So this isn't a company that was founded to do this, but you obviously recognize the need and the applicability
03:07of what you're doing to national security needs.
03:10What can you tell us just about how that pivot has been and what that level of dialogue or interface
03:14with the Pentagon has been like?
03:15Yeah. I mean, I think for us as a company, as you said, you know, we've been in business a
03:19little over 10 years, and we started as a commercial technology and serve many of large global pharma, many financial
03:26services organizations, use our technology for mission-critical AI.
03:29So, you know, we got the call from the DIU, the Defense Innovation Unit, a few years ago, and I've
03:37heard a lot of horror stories about working with the government, right?
03:39Talk about a long timeline.
03:41Long timeline. Contract. Try reading a contract. There's a lot to it.
03:45But there's a moment here that is different.
03:48There's bipartisan support for recognition that we need to enhance America's capabilities in the defense industrial base and in technology.
03:58And so, you know, we thought we had technology that had been proven in the commercial sector that would be
04:05very viable to use in defense.
04:07And thankfully, you know, we've, like I said, we already work on...
04:10Can you say what it was in the commercial sector?
04:12I'm trying to think what you were doing that is now being applied to mind-sweeping.
04:15Well, look, we have a variety of different solutions and use our platform in a variety of different places.
04:21In pharmaceuticals, we help 10 of the top 20 global pharma organizations across their entire pipeline.
04:28So that could be early research and then development all the way through clinical trials and manufacturing.
04:34There's a lot of different parts of that process that could be enhanced by AI.
04:38And in financial services, we have large customers that use us for risk modeling.
04:42You know, you have trillions of dollars in flows in your book, and you want to make sure that risk
04:48is well managed, right?
04:49We'll get to the broader existential question I think a lot of us and companies in AI are facing right
04:54now,
04:54which is what's the involvement of humans going to be going forward?
04:56So we see that in the context of Anthropic, obviously, the debate now, legal case that they have with the
05:01Pentagon.
05:02Their concern about that being ceded to computers entirely.
05:06Where do you see all of this headed?
05:08Obviously, the work that your company is doing is in concert with these unmanned vehicles,
05:11but I imagine there's a human component there as well.
05:13Do you foresee a time when there is no human doing some of this work, or...?
05:19I think that's very hard.
05:21And, look, I've often heard the phrase that, well, China is going to do it.
05:26And, you know, man, what a bad idea, right?
05:29There are lots of things we wouldn't do that China does, right?
05:31And so I think it's a really important debate.
05:36I think in particular with the Pentagon, look, at the end of the day, who sets the ethics?
05:41It is the Pentagon lawyers, it is the Congress, and it's ultimately us as the voters who are putting folks
05:47in place, right?
05:48This is an important conversation we're going to have over the next...
05:50Is it you as an executive as well?
05:51Is it a roiling conversation in the company as well about what the ethical stance of Domino Day Labs is
05:56going to be?
05:57Look, we certainly talk about it, but we're not going to be a vendor who takes on encumbering an organization
06:04with a particular ethical bound.
06:08That's a job for somebody else.
06:09I will say, you know, what I think is critically important for tech companies to do is these technologies have
06:15limitations, as does any technology, right?
06:17And so when you say to the government, I have a piece of technology, you need to define the bounds
06:23in which it can be used the same way.
06:25But are you confident they'll stick to that definition?
06:27Like, you've set out your boundaries.
06:28Do you have faith?
06:29You just said try to read one of these contracts.
06:30Do you have faith that it will be used the way you say it should be used?
06:34Well, look, at the end of the day, that is an imperative of the operators to use the technology in
06:41the right sort of way.
06:42We get this question a lot about the risk of AI.
06:45And I think, ultimately, everyone who's using an AI system still owns the responsibility of what it does.
06:51If you're a pharmaceutical company betting your new drug on AI, are you going to seed that to a machine?
06:56No, right?
06:57And the same thing is true for the operators who use this technology.
07:00So about 30 seconds left.
07:01How quickly were you able to make that pivot?
07:03So you saw this need.
07:04The war is still a very – I mean, it's been eight weeks.
07:07How quickly were you able to make the pivot?
07:09Yeah, so, look, the beautiful thing about our technology, the thing that we've done for the Navy,
07:15is reduce the retraining time from months to days.
07:19And that's a critical imperative when you want to go to a new theater, right?
07:23So the value that we're providing – a technology developed in peacetime over the past four years
07:28is ready to be deployed rapidly, you know, not something that's going to take a year to get deployed.
07:34And thank you very much for being here today.
07:34Fantastic.
07:34Well, yeah.
Comments