- 11 hours ago
Category
🤖
TechTranscript
00:00There is a lot happening right now. We can go through it. The war in Iran and the role that
00:05technology companies have or have not played in it. There's also just a broader discussion about the government's rights to
00:12do with technology that it procures as it wishes. And if it's okay, I'd like to start there.
00:19Palantir has a technology relationship with Anthropic, but it has a technology relationship with many of the Frontier model labs.
00:25And actually, Alex Karp, the CEO, has been relatively clear on what the company's position is. But it's our first
00:33opportunity to talk to you. And so I just asked if you'd reflect on what happened and how Palantir is
00:38approaching its choice of technology partner in light of that.
00:42Well, I think what we want to do, our thesis from the beginning has been that this is a form
00:47of commodity cognition, that actually you're going to have multiple LM providers.
00:52There's going to be a little bit of a horse race, who's better, when they're better. You're going to need
00:55the infrastructure to evaluate which models you should be using and which use cases when.
00:59And that's going to be a moving target. You don't make that determination once, it's going to be ongoing.
01:03So in our context, even in the work that we do with the government, switching even a very complicated, sophisticated
01:08set of workflows is that's order of weeks worth of work.
01:11That's part of the value proposition of what we're doing. And the integration of this stuff to derive economic value
01:17is where we're very focused.
01:19I think in the present moment, we were talking a little bit off camera about the kind of AI doomerism
01:24that sees the country.
01:26I think a lot of that is a consequence of who we're listening to about AI. These revolutions, technology revolution
01:32in general, they're tools-driven revolutions, not concept-driven revolutions.
01:36Galileo did not invent the telescope. He used the telescope to discover the planets in planetary motion.
01:42The power loom, the personal computer, the microscope, the implications of these inventions on humanity were not determined or even
01:50predicted by the inventors.
01:51They were determined by those who wielded the technology itself.
01:54So the people, you know, and the inventors of AI are geniuses. They're national treasures.
02:00But it doesn't actually mean they're right about the implications it's going to have.
02:04The people who I think have really interesting opinions are the frontline factory workers, the ICU nurses, the people whose
02:09jobs are actually being changed by it.
02:11And when you ask them how do they feel about their job with AI, they're actually really positive about it.
02:17But I think more profoundly, when you ask them how do you feel about your children's future in an America
02:22with AI, that's where they're most excited.
02:24And to me, that is the qualitative definition of a thriving middle class, that you believe your children's future will
02:29be better than yours.
02:30So what's Palantir doing on that? I think this might be one of your questions as well, Caroline.
02:34Like, in the general public in the United States, people do have mixed feelings about AI.
02:40Probably they may have no understanding of how AI is impacting the industry they work in.
02:46But, you know, Palantir works across public and private sectors, lots of different types of companies.
02:52Off camera, I gave you the example of General Mills.
02:54Yeah.
02:54It might be a name that the everyday American knows.
02:57Do you feel like you have a duty to go out and give some good PR to AI and its
03:02benefit to society?
03:04The stories that are true and happening.
03:06I mean, he's got a movie project at the moment.
03:08That's what I need us to, yeah.
03:09Yeah, let me tell you about, so a narrative violation here is we work with a submarine parts manufacturer in
03:16the supply chain.
03:16Because they were able to use AI to get rid of the two weeks of planning time where they were
03:20essentially tools down, developing the production plan,
03:23and now AI lets them do that in 10 minutes, they had more work to do, which meant they had
03:27to add a third shift of workers.
03:29You know, so the sort of facile view of this, that somehow it's just linearly going to replace jobs, I
03:35think it's wrong.
03:36Actually, we have a huge amount of human agency in terms of how we apply this technology.
03:41And I think the American people are skeptical of it because mostly what they see is AI slop.
03:46You know, the impact in the economy right now is nascent.
03:51And that's because a lot of the labs are so busy inventing the technology.
03:55Let's call that the first 80% of the work.
03:57Well, there's a second 80% of the work, which is how do you translate this impact into the actual
04:02real economy?
04:03That's the work that we're busy doing every day.
04:05Like, how does this turn into economic value?
04:07And getting those stories out there, I think, is really important.
04:10So what you see is that AI is the antidote to the 20th century's managerial revolution.
04:15I'm going to push back on what you said, that if you talk to the ICU nurse or you talk
04:19to those who are in the factories, that they're excited for their children.
04:22I haven't heard many people say they're excited for their children with AI.
04:26Where are you hearing this?
04:27What's your data?
04:28And how are we going to scale that so that people are excited about it?
04:31Well, there's a question of are you talking to people who are actually using AI versus people who are afraid
04:37of what it's going to do?
04:38So if you go to Tampa General, the ICU nurses at Tampa General, they will give you a very optimistic
04:43view, having experienced how it enables them to spend more time at the bedside of their patient, less time munging
04:50notes and figuring out which patients to spend time with.
04:52So I think the normative value of AI for America is that it should give the American worker superpowers.
04:58How do we make the American worker 50 times more productive than any other worker anywhere else?
05:03And that's going to solve the math problem of how we re-industrialize the country.
05:07We shouldn't be doing that symmetrically with our competitors.
05:10It is David's slingshot against Goliath here.
05:13It's also about training, reskilling, ensuring that the workforce of today can use the tools that they're currently being inundated
05:19with.
05:19And I'm interested, forward deployed engineers is something that we're seeing being used across the board.
05:24And it's something you invented, basically.
05:26That's right.
05:27How is that enabling companies to actually get the RO AI that we've long talked about?
05:32Well, I think it enables you to actually walk many miles in your customer's shoes.
05:36Instead of having some sort of abstract notion of how will this be valuable, you're actually solving backwards from here's
05:42the specific human, here's the work.
05:44How do I build this Iron Man suit around this person?
05:47And what am I learning as a consequence of doing that?
05:49What sort of harnesses are we going to need?
05:51How do we add more value?
05:53What is missing from here?
05:55To the point of the antidote to the managerial revolution, some of the most compelling AI application developers that I'm
06:01seeing are actually in the military.
06:03These are green suitors, uniformed service members without formal computer science degrees.
06:07And what you're seeing is, in general, AI is empowering the vocational worker, empowering the person with specific knowledge as
06:14opposed to just general skills.
06:16And I was, you know, someone who's worked with the military for 20 years, I was asking myself the question,
06:20like, where do these people come from?
06:22And I came to the obvious conclusion, like, they've always been here.
06:26But 10 years ago, what would they have done?
06:28Make a PowerPoint slide?
06:29Try to convince someone else in some program office somewhere their idea was any good?
06:33No.
06:34They're too smart for that.
06:34They know that they would have kind of just been shut down.
06:37Instead, they go off in a corner for a week.
06:39They build it themselves.
06:40And now you're arguing about an empirical reality.
06:42So it's very exciting to see the kind of best ideas from the frontline people have an opportunity to really
06:47succeed and transform these businesses.
06:49Shomar, I appreciate that point.
06:51As you know, Alex Karp and I have done battle over the PowerPoint presentation in the past.
06:56And he's made his views on that very note.
06:58What I think would be very useful in this moment in time is to explain how either specifically the work
07:05that Palant is doing or just industry in general and how it's changing the battlefield.
07:10There is a war in Iran.
07:12And, you know, there was reporting about the planning of that and America's ability to move quickly.
07:18And while we've not yet had a chance to discuss it over the last six months, you know, you yourself
07:23have an attachment to the U.S. military that's relatively new.
07:27Just reflect on that, please.
07:29Yeah, I think obviously current operations aren't going.
07:33But I think people will reflect back and say this is the first large scale combat operation that was really
07:39driven, enhanced, made substantially more productive with technology, with AI.
07:44Yes.
07:44And, you know, the word targeting, I think, to the lay public has some sort of connotation.
07:49It sounds like someone's shooting.
07:50But it's actually a very bureaucratic, large scale doctrinal process.
07:54So if you thought about that like any, you know, I have a value chain to manage.
07:57I have a production process to manage.
07:59I have a targeting process to manage.
08:01Being able to accelerate that.
08:02So if you think about Gulf War I, I think, or Gulf War II, sorry.
08:08Gulf War II, we did about 1,000 targets.
08:10It was six months of planning for roughly 50 to 100 people.
08:14And in this conflict, you're looking at that equivalent of work for twice as many targets was done by one
08:20person in two weeks.
08:21So how do we give our service members that Iron Man suit?
08:24We're making them 50 times more productive than the adversary.
08:26You say Iron Man suit.
08:27Yeah, the conceptual Iron Man suit, right?
08:29Like, how do I make them superhuman?
08:31And this is a realization of the third offset strategy.
08:34You know, the first offset was nuclear weapons.
08:36The second offset was precision guided munitions.
08:38How do we, we're not going to, we're not going to, we don't want to engage symmetrically with the Soviets.
08:43Like, every missile they have, we need to have a missile.
08:45No, we're going to have more precise missiles.
08:47So we'll need less of them as a consequence.
08:49Don't forget, in World War II, only 6% of bombs dropped hit their targets.
08:52But this is very asymmetric right now.
08:54And that is exactly the point.
08:56That is deterrence.
08:56So, you know, I just wrote a book that came out and mobilized.
08:58The whole point of the book is we have been in a frog boil.
09:01That really, we have lost deterrence over the last 12 years or so.
09:04We had the annexation of Crimea in 14, the militarization of the Spratly Islands in 15,
09:09Iran with breakout capability to get the bomb in 17.
09:12We've had a pogrom in Israel.
09:13We have the Houthis holding trade hostage in the Red Sea, skirmishes India, Pakistan.
09:20You can look at this and you can say, like, maybe we're going to look back in 10 years and
09:23say,
09:23World War III had already started and we didn't know it.
09:25So if we want to prevent that from happening, what do we need to do right now?
09:28We need to mobilize.
09:29We need to get very serious about the magazine depth we have, our ability to make things,
09:34produce things in this country.
09:35There's the obvious part of building weapons to deter our adversaries.
09:38If you only have eight days of weapons on hand to fight China, that's not very scary.
09:42You know, you need 800 days.
09:44But I think more profoundly it's protecting our supply chains.
09:48You know, the Strait of Hormuz is a great example of that.
Comments