WEBVTT 00:00:00.000 --> 00:00:02.997 It's a risk our next guest has been warning the tech industry about for 00:00:03.080 --> 00:00:06.797 months. His company writing in June, "Your AI sovereignty dictates your 00:00:06.880 --> 00:00:11.277 institution's future." Here with us now at Post 9 in a CNBC exclusive interview 00:00:11.360 --> 00:00:14.757 with Palantir [music] co-founder and CEO Alex Karp. Welcome back. It's good to 00:00:14.840 --> 00:00:16.277 see you. >> Delighted to be here. You know, I've 00:00:16.360 --> 00:00:20.877 been here many times with you. And you guys are tough, but among my favorites. 00:00:20.960 --> 00:00:24.837 >> Okay, well, thank you. >> uh >> Well, we're excited to talk to you 00:00:24.920 --> 00:00:26.917 because this is a really important week. 00:00:27.000 --> 00:00:32.877 And there's been this explosion of AI safety warnings. Where do you stand on 00:00:32.960 --> 00:00:37.917 all of it? >> so okay, Palantir we've been we've been in the forefront of using AI 00:00:38.000 --> 00:00:41.557 in the military context, the first company to build actually a functional 00:00:41.640 --> 00:00:45.717 AI technology. As is currently well known, we're dominating the application 00:00:45.800 --> 00:00:49.837 space. We have a product. So, there's a lot of we have a lot of experience with 00:00:49.920 --> 00:00:54.117 like what works, what doesn't. There are really different kinds of safety issues, 00:00:54.200 --> 00:00:56.237 quite frankly. Um 00:00:56.320 --> 00:00:59.557 almost all my we dominate US commercial 00:00:59.640 --> 00:01:04.557 as a company, and the clients, our partners, are livid at 00:01:04.640 --> 00:01:08.957 this kind of safety they're most upset about. What they believe is the theft of 00:01:09.040 --> 00:01:13.597 their alpha, meaning their ideas about how to run a business, their data, it 00:01:13.680 --> 00:01:17.517 landing in a model, and they find out that the the competitor next door has 00:01:17.600 --> 00:01:22.037 all their alpha, meaning how they do their business, and they paid for it. I mean, 00:01:22.120 --> 00:01:26.117 I can't repeat what the customers are saying, but they're basically like one 00:01:26.200 --> 00:01:30.037 to give you the one that was the least vulgar, it's like, but 00:01:30.120 --> 00:01:34.197 but how can they tell me not to wear a condom? Like, this is This is And that's 00:01:34.280 --> 00:01:38.557 the least vulgar person that we deal with. They're like livid. And so, 00:01:38.640 --> 00:01:41.597 they're livid. So, that's >> Wait, what? Can you just like translate 00:01:41.680 --> 00:01:44.957 what you're talking about? So, is this because of the LLM's fault they're sharing 00:01:45.040 --> 00:01:47.757 >> No, no. Open The reason why the sovereignty 00:01:47.840 --> 00:01:51.037 revolution, which we helped launch 00:01:51.120 --> 00:01:56.117 together with our our partner Jensen Huang, the in 00:01:56.200 --> 00:01:57.357 famous 00:01:57.440 --> 00:02:00.677 founder and head of Nvidia is uh um 00:02:00.760 --> 00:02:05.677 the when you use a closed model and you pay for tokens, you're not The reason 00:02:05.760 --> 00:02:08.677 the tokens are discounted is not so they can grow the market. It They're 00:02:08.760 --> 00:02:12.397 discounted so that they can get access to your IP so that their models get 00:02:12.480 --> 00:02:16.997 better. And there was functionally no honest conversation about this or 00:02:17.080 --> 00:02:21.317 disclosure. Now, that's point one. Point two is without an application layer and 00:02:21.400 --> 00:02:25.757 without post training, which is currently being done with open 00:02:25.840 --> 00:02:29.317 way models, you in the classified side and it's just 00:02:29.400 --> 00:02:31.077 taking off across America cuz it's 00:02:31.160 --> 00:02:33.317 cheaper and ends up being safer and you 00:02:33.400 --> 00:02:37.317 control your future and your present. Um 00:02:37.400 --> 00:02:41.517 you are de facto outsourcing the value in there. By the way, it's very much 00:02:41.600 --> 00:02:45.477 linked to the safety concerns. Now, Dario is one of the most important 00:02:45.560 --> 00:02:48.597 entrepreneurs. I've been at this for a long time. I've never seen anyone come 00:02:48.680 --> 00:02:49.677 from the back >> you like him. 00:02:49.760 --> 00:02:53.797 >> I I have a lot of respect for him. And I have a lot of respect for him because 00:02:53.880 --> 00:02:59.037 he's actually believes of course he's a business person to most sometimes. So, 00:02:59.120 --> 00:03:02.317 of course when the company's not doing as well, these things come forward and 00:03:02.400 --> 00:03:05.637 then they talk about safety and when it's doing better, they they they 00:03:05.720 --> 00:03:09.157 I I I don't think that's actually what's driving this, but you know, I mean it 00:03:09.240 --> 00:03:13.797 does happen regularly. But I think the you know, he's what makes him so strong 00:03:13.880 --> 00:03:15.957 is he's ethical and a business person. 00:03:16.040 --> 00:03:20.557 Okay, so um but but but what's actually happening is 00:03:20.640 --> 00:03:23.637 very different than what people are reporting. Everyone thinks this is a 00:03:23.720 --> 00:03:28.397 safety versus non-safety versus regulation. What What What he I believe 00:03:28.480 --> 00:03:32.797 is realized is it if you if take the 00:03:32.880 --> 00:03:36.757 a philosophical frame they have which is we are deserving of having all the IP in 00:03:36.840 --> 00:03:40.357 the world, we the thousand people who think the way we do, 00:03:40.440 --> 00:03:43.957 uh that's why you can't work with the military. That's why you can take the IP 00:03:44.040 --> 00:03:46.397 from businesses and put it in your model. 00:03:46.480 --> 00:03:50.637 That That model would require severe 00:03:50.720 --> 00:03:54.197 liability protection. And there's only one institution that 00:03:54.280 --> 00:03:57.877 can do that, the US government. And then that's before you get to the danger. So, 00:03:57.960 --> 00:04:02.077 what was really happening is the investors loved when he was 00:04:02.160 --> 00:04:05.077 migrating all the value of everyone else's business to their business cuz 00:04:05.160 --> 00:04:09.037 they're investors. They don't realize it's like it's like 00:04:09.120 --> 00:04:13.517 every mark. They don't realize they're the mark because now you first you 00:04:13.600 --> 00:04:17.077 migrate all the IP to your business and then because you would need uncap you 00:04:17.160 --> 00:04:20.717 need cap liability both vis-a-vis the business and 00:04:20.800 --> 00:04:23.917 vis-a-vis the dangers, you've got to migrate the business to the government. 00:04:24.000 --> 00:04:27.437 It has to be nationalized. So, what's actually happening is this is reported 00:04:27.520 --> 00:04:29.317 as regulation non-regulation movement. 00:04:29.400 --> 00:04:33.157 That's actually not what this movement is. It's like the view that I believe 00:04:33.240 --> 00:04:37.037 they have is these businesses have to be nationalized because if you don't 00:04:37.120 --> 00:04:40.557 nationalize them, every single one of my clients is going to sue. 00:04:40.640 --> 00:04:42.037 >> That you think that's what they want? 00:04:42.120 --> 00:04:44.397 >> Well, obviously, like if you have a 00:04:44.480 --> 00:04:48.477 business that presupposes unlimited risk 00:04:48.560 --> 00:04:52.557 both vis-a-vis liability risk vis-a-vis the clients cuz I don't think any of our 00:04:52.640 --> 00:04:54.477 clients realize their IP was migrating. 00:04:54.560 --> 00:04:57.717 Vis-a-vis the world because you're saying what is the liability risk for 00:04:57.800 --> 00:05:02.317 destroying 10% of the company? 10% of the world. It's discounted into the 00:05:02.400 --> 00:05:05.677 present. Okay, but there's only one institution. And by the way, that's not 00:05:05.760 --> 00:05:08.877 actually what's happening. They want three institutions. It's going to be 00:05:08.960 --> 00:05:11.637 like Europe, Canada, and America. We 00:05:11.720 --> 00:05:16.397 migrate all the IP of America from our businesses to this institution. Then the 00:05:16.480 --> 00:05:20.077 institution is going to get full liability protection because it's nationalized. 00:05:20.160 --> 00:05:21.957 >> think Open AI wants this and what about this? 00:05:22.040 --> 00:05:25.557 >> very No, I think I think these are very different things, but I think that is 00:05:25.640 --> 00:05:30.237 the core philosophy of the EA movement imposed upon a business. 00:05:30.320 --> 00:05:33.957 >> EA effective altruism. >> so that's actually what is driving this. 00:05:34.040 --> 00:05:37.237 Now, of course, the other model companies have very different I think 00:05:37.320 --> 00:05:40.557 what most what what they want is and I think is we first we should 00:05:40.640 --> 00:05:43.437 self-regulate. If you do not self-responsibly self-regulate. By the 00:05:43.520 --> 00:05:47.757 way, though, the biggest danger currently of this revolution is that it 00:05:47.840 --> 00:05:51.477 generates inescapably I think it will raise the 00:05:51.560 --> 00:05:55.437 votes of most people but you can't have a revolution where I become 50 times 00:05:55.520 --> 00:05:58.117 wealthier and everyone else becomes 10% 00:05:58.200 --> 00:06:02.517 wealthier. It's going to drive political insanity. Political insanity, what does 00:06:02.600 --> 00:06:06.877 that look like? It looks like obviously wealthy people should pay more than fair 00:06:06.960 --> 00:06:11.957 their their share of the taxes but a wealth tax kills entrepreneurship. It 00:06:12.040 --> 00:06:14.917 actually is it by the way it's massively beneficial to me. 00:06:15.000 --> 00:06:15.917 >> How do you share the benefits? 00:06:16.000 --> 00:06:19.797 >> So you have to find ways to actually tax at the high end that that aren't a 00:06:19.880 --> 00:06:23.917 wealth tax. There are ways to do this. I think the way you do it is you tie 00:06:24.000 --> 00:06:26.797 reduction of waste fraud and abuse to 00:06:26.880 --> 00:06:29.237 increase taxes among the very wealthy. 00:06:29.320 --> 00:06:33.157 So but in any case you you have to you you the other political insanity and 00:06:33.240 --> 00:06:36.197 then realize the financial thing is look at the far right and far left in this 00:06:36.280 --> 00:06:40.797 country. This is directly tied to the fact that many people are like screw 00:06:40.880 --> 00:06:43.997 this. I don't want these people becoming a hundred times wealthier. I'm going to 00:06:44.080 --> 00:06:47.317 blow up the whole system. What does that blowing up the whole system looks like? 00:06:47.400 --> 00:06:51.277 It looks like DSA people who are very happy to talk about how they're not 00:06:51.360 --> 00:06:53.797 anti-Jewish. They love that. They don't like to talk about how they're not 00:06:53.880 --> 00:06:57.637 anti-black cuz that's something they hide. You have the far right. It's like 00:06:57.720 --> 00:07:01.237 everything is a data center run by somebody. It's like they attack us for 00:07:01.320 --> 00:07:04.837 running data. We don't even build data centers. [laughter] We get attacked. So 00:07:04.920 --> 00:07:08.677 that is like and by the way the one of the biggest mistakes the labs 00:07:08.760 --> 00:07:13.437 have made is we are in competition with China. Some people think they're 00:07:13.520 --> 00:07:17.037 enemies. Some people think they're competitors. Pick your spectrum. 00:07:17.120 --> 00:07:21.437 Either we are going to win and provide the stack to the world and stack doesn't 00:07:21.520 --> 00:07:25.317 mean you have to take the whole part or if they're going to win. And that means 00:07:25.400 --> 00:07:29.357 the the danger is not just a danger of dangerous things happening to humanity 00:07:29.440 --> 00:07:33.957 which by the way what business isn't responsible for minding your own safety? 00:07:34.040 --> 00:07:37.317 You can't go to a what you know another reason our customers are livid is like 00:07:37.400 --> 00:07:39.517 they build chemicals >> a whole new frontier, right? When 00:07:39.600 --> 00:07:41.637 they're talking about the safety of 00:07:41.720 --> 00:07:44.397 humanity and creating a technology and 00:07:44.480 --> 00:07:47.517 AI that can break free this misalignment 00:07:47.600 --> 00:07:52.197 from human intention and then do really bad stuff to us. I mean, that's a 00:07:52.280 --> 00:07:56.357 societal issue, isn't it? >> Of course it's a societal issue, but you 00:07:56.440 --> 00:07:57.717 >> You think they can self-regulate that? 00:07:57.800 --> 00:08:01.477 >> Well, you have to find a way to set reasonable guidelines that are enforced, 00:08:01.560 --> 00:08:05.397 but you can't do it. The first line of defense is you're liable for your own 00:08:05.480 --> 00:08:07.957 actions. So, it's it's a weird twist. 00:08:08.040 --> 00:08:11.917 They're asking for societal regulation to get out of the first line of defense, 00:08:12.000 --> 00:08:16.597 which is if you build a technology that can destroy 10% of the of of of the 00:08:16.680 --> 00:08:20.557 world. That is a that has civil and criminal 00:08:20.640 --> 00:08:24.677 liability attached to it. The first step is to say, "Okay, it is not The first 00:08:24.760 --> 00:08:27.517 step isn't to say we get out of it because we want to self-regulate or we 00:08:27.600 --> 00:08:31.437 want regulation." The first step is to say, "Okay, you've disclosed this. What 00:08:31.520 --> 00:08:32.557 are you doing about it?" >> Right. 00:08:32.640 --> 00:08:35.397 >> And the second step is to say, "If you're not doing any you're not being 00:08:35.480 --> 00:08:38.237 responsible, we're we're going to have to get people involved." That's going to 00:08:38.320 --> 00:08:41.757 be also very complicated because regulating these things need people. You 00:08:41.840 --> 00:08:45.357 need people who understand them. Another problem we have is seemingly everyone 00:08:45.440 --> 00:08:47.237 who understands this is on some payroll. 00:08:47.320 --> 00:08:49.837 So, it's like, you know, it's like who you're going to trust? You know, when we 00:08:49.920 --> 00:08:51.597 built nuclear energy, we built the bomb. 00:08:51.680 --> 00:08:53.117 People did it because they believed. 00:08:53.200 --> 00:08:56.957 They were not on the payroll. >> Um this the liability that you're 00:08:57.040 --> 00:09:00.797 talking about, how does that get written into the S-1? What do risk factors sound 00:09:00.880 --> 00:09:04.997 like? >> This is the thing. You're assuming that 00:09:05.080 --> 00:09:09.237 there will be an S-1. Now, I don't Okay, maybe there will be. The only way to 00:09:09.320 --> 00:09:12.637 deal with this kind of liability is to go to the government and say, 00:09:12.720 --> 00:09:16.477 "Nationalize us, please." You know, a lot of times in business people don't 00:09:16.560 --> 00:09:20.277 say what they want. They're leading you to the conclusion. The reason why people 00:09:20.360 --> 00:09:24.037 don't have not understood this is again, every investor, and there are a lot of 00:09:24.120 --> 00:09:27.117 people who think they're going to make money. They're like, "Oh, the businesses 00:09:27.200 --> 00:09:31.317 are the mark." What is the first lesson in poker and in 00:09:31.400 --> 00:09:33.917 life? If you you you think the other 00:09:34.000 --> 00:09:36.517 person's the mark, you're the mark. 00:09:36.600 --> 00:09:40.677 You are the mark. And it will be nationalized simply 00:09:40.760 --> 00:09:43.677 because of the liability issues they're talking about. 00:09:43.760 --> 00:09:45.837 >> the Trump administration's going to get on board with that? 00:09:45.920 --> 00:09:48.477 >> I don't know what they're going to do, and I think that's probably a barrier to 00:09:48.560 --> 00:09:53.117 entry, but it it's like you can delay it. You have to There's only one When 00:09:53.200 --> 00:09:56.477 you have unlimited liability, the only way to deal with it would be to go to 00:09:56.560 --> 00:09:59.597 the government and say, "We'll give you 50% of our business." Of course, the 00:09:59.680 --> 00:10:02.317 value of the business will collapse first when you do that, collapse another 00:10:02.400 --> 00:10:06.197 time when you have people on the board, collapse another time. All of us might 00:10:06.280 --> 00:10:09.357 even collapse the whole market, but you That is the way you essentially But 00:10:09.440 --> 00:10:12.957 again, you have to look at the philosophy of it. Again, this is where 00:10:13.040 --> 00:10:16.917 the investors have just made such a big mistake with their greed. Like, a little 00:10:17.000 --> 00:10:21.117 bit of greed is great, but like it's like No, they didn't mind when the when 00:10:21.200 --> 00:10:23.357 the when the IP is being migrated from 00:10:23.440 --> 00:10:25.997 every business to an entity they own. 00:10:26.080 --> 00:10:30.557 But the next step is obviously to take that value and migrate it to the 00:10:30.640 --> 00:10:33.637 government. But oh, that would never happen to me. 00:10:33.720 --> 00:10:37.117 And like And again, this is where I have tremendous respect for people like that 00:10:37.200 --> 00:10:41.237 because he's thinking ethically according to his ethical frame. He's not 00:10:41.320 --> 00:10:44.917 thinking primarily financially. This is what's made him such a great business 00:10:45.000 --> 00:10:48.077 leader. This is the reason why the best engineers 00:10:48.160 --> 00:10:52.197 researchers want to work for him. It's the reason why he came from fifth behind 00:10:52.280 --> 00:10:54.917 to number one. This is a very, very 00:10:55.000 --> 00:10:57.357 smart, shrewd, ethical person. The 00:10:57.440 --> 00:11:02.237 ethical frame is different than the frame most of us have, which is we need 00:11:02.320 --> 00:11:04.277 to share wealth, we need to expand the 00:11:04.360 --> 00:11:06.557 wealth, we need to diverse put more 00:11:06.640 --> 00:11:09.637 diversity in terms of who owns and who 00:11:09.720 --> 00:11:12.517 regulates and who controls and 00:11:12.600 --> 00:11:16.877 if it started in America, it could have only been built in America, only America 00:11:16.960 --> 00:11:19.597 would have allowed you to buy it, only America provided the regulatory 00:11:19.680 --> 00:11:21.597 framework where this could have happened. 00:11:21.680 --> 00:11:26.397 America should be the final arbiter of it in the military and in regulation, 00:11:26.480 --> 00:11:28.557 and that's exactly what they do not want. 00:11:28.640 --> 00:11:32.757 >> It sounds like you think his examples of potential harm, right, disabling the 00:11:32.840 --> 00:11:35.317 internet, You think those are in the realm of the possible. You're not 00:11:35.400 --> 00:11:38.317 disputing any of that. >> I I'm I'm What I'm saying is if I let's 00:11:38.400 --> 00:11:41.517 just take nuclear energy. There it's very like for example, one of 00:11:41.600 --> 00:11:45.077 the reasons France is still much stronger than you'd expect is they moved 00:11:45.160 --> 00:11:47.437 to nuclear. If you look interestingly, 00:11:47.520 --> 00:11:49.157 if you look at 00:11:49.240 --> 00:11:52.197 energy expenditure per unit per person 00:11:52.280 --> 00:11:54.597 as a function of GDP, it's America, 00:11:54.680 --> 00:11:58.037 China, uh Israel, and then Switzerland, 00:11:58.120 --> 00:12:00.517 roughly France. Okay. So, this is very 00:12:00.600 --> 00:12:03.037 This is technology that is sharp. It's 00:12:03.120 --> 00:12:05.997 used in medicine. It's used in nuclear. 00:12:06.080 --> 00:12:08.837 It powers our deterrents above the air, 00:12:08.920 --> 00:12:12.357 below the air, in the ocean. Okay. 00:12:12.440 --> 00:12:16.517 It It can also be abused, and there are also accidents. 00:12:16.600 --> 00:12:20.637 The first step, you cannot say I'm building something, but I have no idea 00:12:20.720 --> 00:12:23.517 how to bound it. And I don't believe that's what But again, that's I don't 00:12:23.600 --> 00:12:24.877 believe that's what they're saying. 00:12:24.960 --> 00:12:25.757 >> That's not what they're saying. 00:12:25.840 --> 00:12:28.237 >> No. No. No. That's what they're saying, but that's not what they mean. 00:12:28.320 --> 00:12:30.957 >> Right. >> What they mean is the obvious solution 00:12:31.040 --> 00:12:35.197 is America America gives a risk 00:12:35.280 --> 00:12:39.037 essentially limits the risk in return for owning it. Or a national or 00:12:39.120 --> 00:12:42.437 multinational government does. That's what they actually mean. That's not what 00:12:42.520 --> 00:12:45.797 they're saying. Because you couldn't come out to say it cuz somebody has It's 00:12:45.880 --> 00:12:48.917 like two It's like the investors are going to be what the F? I thought 00:12:49.000 --> 00:12:50.757 someone else was the mark. >> [laughter] >> Right. 00:12:50.840 --> 00:12:52.997 >> I'm counting my huge pile of pennies. 00:12:53.080 --> 00:12:56.397 >> for marketing an IPO. >> Well, it Yeah, I mean, you know, people 00:12:56.480 --> 00:12:57.877 don't like finding out they're the mark. 00:12:57.960 --> 00:13:02.557 But again, the way you The way you can avoid being in a mark in in life is to 00:13:02.640 --> 00:13:06.317 assume when someone is pointing out someone else is the sucker, you're the 00:13:06.400 --> 00:13:09.037 sucker. >> What So, you have I mean, for people 00:13:09.120 --> 00:13:13.357 don't realize like Palantir, you're kind of the the in between, right? Those 00:13:13.440 --> 00:13:17.557 frontier models and then the enterprise >> You can think of it as you There's 00:13:17.640 --> 00:13:19.077 something called an application layer. 00:13:19.160 --> 00:13:22.757 Historically, the reason we thrived even before people realized the safety issue 00:13:22.840 --> 00:13:25.517 was this big was that to make these 00:13:25.600 --> 00:13:28.997 models actually work precisely over very 00:13:29.080 --> 00:13:35.197 complex use cases, think building complicated manufacturing, Nvidia's 00:13:35.280 --> 00:13:38.677 >> at Chipotle. >> Food safety, uh military, Nvidia's 00:13:38.760 --> 00:13:42.757 supply chain, things that are very proprietary and need precision where you 00:13:42.840 --> 00:13:45.917 own the precision, you need an application 00:13:46.000 --> 00:13:49.837 layer. And the application layer also allows you to use models that might 00:13:49.920 --> 00:13:53.957 otherwise not be safe, like closed models that may in fact be migrating 00:13:54.040 --> 00:13:56.157 your IP to them. >> Right. So, what I was going to ask is 00:13:56.240 --> 00:13:58.637 you do a lot of this for the federal government and for the defense 00:13:58.720 --> 00:14:02.677 department. And so, you have experience in >> Well, our we are the single most 00:14:02.760 --> 00:14:06.597 enterprise software company, I believe, in the world. In the US, it grew 149%. 00:14:06.680 --> 00:14:10.717 We are famous for what we do on the battlefield. And on the battlefield, you 00:14:10.800 --> 00:14:13.837 have exactly these issues. >> Exactly. That's where I'm going. Like, 00:14:13.920 --> 00:14:16.757 you can't make these safety mistakes when you're talking about using AI in 00:14:16.840 --> 00:14:20.197 the workforce. >> you have you Look, you you you're always in the 00:14:20.280 --> 00:14:23.037 business of reducing things beyond what 00:14:23.120 --> 00:14:27.397 we could do yesterday in safety and in positive and negative. The thing that's 00:14:27.480 --> 00:14:30.397 very important about the battlefield base, 00:14:30.480 --> 00:14:33.477 first of all, you are you 00:14:33.560 --> 00:14:37.797 may or not may not You can help bring our men and women home. So, there's like 00:14:37.880 --> 00:14:41.677 the safety thing. Then, there's the precision of what happens. And then, 00:14:41.760 --> 00:14:44.717 very interestingly, national security is different 00:14:44.800 --> 00:14:47.917 technically than people think. There's like even inside an enterprise, there 00:14:48.000 --> 00:14:51.557 are multiple levels of security. So, like, for example, you have levels of 00:14:51.640 --> 00:14:54.397 classification. But then, there are classifications that you need to be 00:14:54.480 --> 00:14:57.557 classified to understand. So, when you're on the battlefield, you're 00:14:57.640 --> 00:14:59.597 working at when the when a missile lands 00:14:59.680 --> 00:15:04.637 on our enemy's head, it actually is the concatenation of data sets and precision 00:15:04.720 --> 00:15:08.357 that's owned in our anthology and in the model they use, where no single 00:15:08.440 --> 00:15:11.757 component would be able to see the other component. So, it ends up being But by 00:15:11.840 --> 00:15:15.397 the way, you have very similar issues with like a supply chain at Nvidia or 00:15:15.480 --> 00:15:17.357 food safety >> what I'm wondering is how do you make 00:15:17.440 --> 00:15:21.997 sure that that's determined by a human and not by the AI. Cuz that's some of 00:15:22.080 --> 00:15:24.957 the Some some say it's science fiction, but that's what some of the worry is. 00:15:25.040 --> 00:15:27.357 >> the human in the loop thing is a very, 00:15:27.440 --> 00:15:29.837 very big issue. We have powered human in 00:15:29.920 --> 00:15:32.437 the loop. The com- complication in 00:15:32.520 --> 00:15:35.077 defense is you have different ethical 00:15:35.160 --> 00:15:39.957 standards for offense and defense. So, like if I'm attacking you, I have to do 00:15:40.040 --> 00:15:43.757 I'm only allowed to do X things if you're attacking me. So, the problem in 00:15:43.840 --> 00:15:47.957 this context is if someone is if if like we're peer-to-peer adversary, you're 00:15:48.040 --> 00:15:51.797 going to have to have a response that is not a human in the loop. It's so quick. 00:15:51.880 --> 00:15:55.677 And and also with this kind of technology, you can often predict. So, 00:15:55.760 --> 00:15:58.557 if we can predict you're going to you're going to attack me, is that offense or 00:15:58.640 --> 00:16:02.357 defense? Because in offense, you definitely need a human in the loop. In 00:16:02.440 --> 00:16:04.277 defense, there's going to be no time. 00:16:04.360 --> 00:16:08.397 Okay, what if I can predict in in 35 seconds or in 3 hours you're going to 00:16:08.480 --> 00:16:10.357 attack me? Is that offense or defense? 00:16:10.440 --> 00:16:14.877 And so, there there there a lot of The reason why you really, unfortunately, 00:16:14.960 --> 00:16:17.997 many of my academic colleagues don't want to hear this, you really have to 00:16:18.080 --> 00:16:23.237 pick a side because in the end, America in all its greatness is not is a 00:16:23.320 --> 00:16:26.277 country and we're not perfect, but we are great. 00:16:26.360 --> 00:16:29.477 It is a decision. And by the way, I've spent half my life studying Chinese 00:16:29.560 --> 00:16:33.277 culture. I I'm a very high-level practitioner of of Tai Chi, one of the 00:16:33.360 --> 00:16:36.597 four great treasures. I mean, no way is it xenophobic toward China. I I have 00:16:36.680 --> 00:16:40.117 great admiration for it's like the one of the oldest, most productive cultures 00:16:40.200 --> 00:16:43.877 in the world. I mean, I can give you long lectures on this, but in the end, 00:16:43.960 --> 00:16:46.637 there's a choice. Europe has decided to 00:16:46.720 --> 00:16:49.437 not participate. Israel was small. 00:16:49.520 --> 00:16:53.157 Where is a choice? Is it us making these very hard decisions that will be 00:16:53.240 --> 00:16:56.317 imperfect, or is it them? So, next week 00:16:56.400 --> 00:16:58.517 when she and the president meet, can 00:16:58.600 --> 00:17:00.677 they agree on some shared risks? I I 00:17:00.760 --> 00:17:03.397 mean, I would love it if it happened. I 00:17:03.480 --> 00:17:06.277 think it is unlikely we're going to The 00:17:06.360 --> 00:17:08.397 problem is we're we're we're tight 00:17:08.480 --> 00:17:13.557 enough as competitors that both sides believe they could win. And so, I I 00:17:13.640 --> 00:17:15.277 would Look, I would love it to happen. 00:17:15.360 --> 00:17:18.997 I'm not in that room. I I would If it happened that we'd all be happier 00:17:19.080 --> 00:17:22.477 Palantir we'd be happier. Um if you could have some kind of 00:17:22.560 --> 00:17:25.757 framework where the two dominant countries could control it that would 00:17:25.840 --> 00:17:29.077 change the world. I would be surprised if it happened. 00:17:29.160 --> 00:17:32.957 >> I was going to ask your perspective on on some of these on the wars. I mean you 00:17:33.040 --> 00:17:36.237 have said and you have said we have the best technology and we have the we're 00:17:36.320 --> 00:17:39.517 winning the AI race and we have companies like Palantir working on it 00:17:39.600 --> 00:17:44.317 and yet you know Ukraine war, Iran war we're kind of in these in these stuck 00:17:44.400 --> 00:17:46.837 places right now. >> Um you know when you're talking about 00:17:46.920 --> 00:17:50.117 battlefield performance where the Ukraine's a good example. 00:17:50.200 --> 00:17:53.757 Every day when I got there day five there was not a single person who didn't 00:17:53.840 --> 00:17:55.357 think they were going to lose. 00:17:55.440 --> 00:17:59.717 Like so you have always have to look what what happened against expectations 00:17:59.800 --> 00:18:03.157 and that the countries that are in the west that are are are wield these 00:18:03.240 --> 00:18:07.597 technologies are Ukraine, Israel and America. I guarantee you if you got a 00:18:07.680 --> 00:18:10.077 classified report of from the from China 00:18:10.160 --> 00:18:12.797 or Russia one of their biggest concerns 00:18:12.880 --> 00:18:17.317 would be the systemic outperformance of these countries on the battlefield where 00:18:17.400 --> 00:18:21.357 these technologies have been implemented. Guarantee. In fact, you 00:18:21.440 --> 00:18:24.877 know, as far as is possible for me to know that without knowing it cuz I don't 00:18:24.960 --> 00:18:27.237 interact with those countries I can tell you this is true. 00:18:27.320 --> 00:18:31.277 >> You think that deters them from Taiwan or something else? 00:18:31.360 --> 00:18:34.317 >> Look, there are a lot of variables that go into these decisions. 00:18:34.400 --> 00:18:37.677 They they the the diagnostic of America 00:18:37.760 --> 00:18:42.077 and its capabilities on the battlefield it means that if they believe we were 00:18:42.160 --> 00:18:45.997 actually to use them it's just a much higher threshold. But you know, I mean 00:18:46.080 --> 00:18:50.197 like these things are not as like these are multi-variable 00:18:50.280 --> 00:18:54.157 equations and also quite frankly often have to do with domestic things that we 00:18:54.240 --> 00:18:57.477 often overlook. Like where when you attack may 00:18:57.560 --> 00:19:01.397 just have to do with like what is your stature at home? 00:19:01.480 --> 00:19:03.757 >> So is there anything if you saw it in in 00:19:03.840 --> 00:19:07.477 the models that they were doing in these AI companies where you would where that 00:19:07.560 --> 00:19:10.837 would that would freak you out and say we can't like 00:19:10.920 --> 00:19:12.677 we have to draw the line and we can't use this. 00:19:12.760 --> 00:19:16.437 >> It's it's it's it's this is a situation where we all need to First of all, we 00:19:16.520 --> 00:19:20.677 need to get serious. You know, we're talking about like we 00:19:20.760 --> 00:19:23.877 like I'm always telling people in different contexts, you know, there is a 00:19:23.960 --> 00:19:27.277 rule here for every every single business person looking at the show, 00:19:27.360 --> 00:19:31.357 every single business that's listed or not listed has a set of rules. It's 00:19:31.440 --> 00:19:35.757 called I deliver products that are where I do not deliver them if I believe in it 00:19:35.840 --> 00:19:39.677 to be harmful. You just have to start there. And again, I I think the 00:19:39.760 --> 00:19:43.237 discussion that we're having is not what this is about. the discussion we're 00:19:43.320 --> 00:19:45.797 having is I want to do things. I want 00:19:45.880 --> 00:19:47.997 new rules for myself and I don't want to 00:19:48.080 --> 00:19:50.677 pay the cost true cost of that, which is 00:19:50.760 --> 00:19:55.437 caught liability, primarily civil, but potentially obviously criminal. And that 00:19:55.520 --> 00:19:59.157 every single This is why the business people I mean, I talk to business people 00:19:59.240 --> 00:20:02.717 every day. I've never seen them this angry. Like 00:20:02.800 --> 00:20:06.797 it's like and every conversation's like this. Wait a minute. They got my IP. 00:20:06.880 --> 00:20:10.477 They want to play by rules I don't play by, whether that's a hospital service, 00:20:10.560 --> 00:20:13.397 doctors, real estate. Every single 00:20:13.480 --> 00:20:16.077 business person in the country worries 00:20:16.160 --> 00:20:20.597 every day about doing something that they didn't even know was harmful. 00:20:20.680 --> 00:20:24.597 Forget knowing it was harmful. You You can ruin your business with something I 00:20:24.680 --> 00:20:28.237 know it's crazy. Like you find out there's something in your product you 00:20:28.320 --> 00:20:32.557 had you did your absolute best to keep out of it out of the product. You you 00:20:32.640 --> 00:20:36.037 still can be found liable. Like you think the plaintiff's attorneys are 00:20:36.120 --> 00:20:38.797 running up there saying, oh, oh, gee whiz. 00:20:38.880 --> 00:20:41.237 >> You didn't mean it. You didn't mean it. 00:20:41.320 --> 00:20:44.117 >> No, you're paying your billions of dollars. 00:20:44.200 --> 00:20:47.757 You lose your house. You lose your future. 00:20:47.840 --> 00:20:51.037 >> So they face lot So they >> No, I'm just I'm just saying you can't 00:20:51.120 --> 00:20:54.277 change the rule book, but but again, I don't actually 00:20:54.360 --> 00:20:58.717 do not underestimate Dario's talent. He That's not That's not what he's actually 00:20:58.800 --> 00:21:02.797 saying. What he's actually saying is this technology is so unique, it can 00:21:02.880 --> 00:21:05.677 only be harnessed in a multi 00:21:05.760 --> 00:21:07.237 jurisdictional, meaning more than one 00:21:07.320 --> 00:21:10.597 country, and I can only do it if you if 00:21:10.680 --> 00:21:15.317 I get to transfer all liability concerns within a million up to a million dollars 00:21:15.400 --> 00:21:18.557 to the US government and probably European and Canadian governments will 00:21:18.640 --> 00:21:20.877 be the fantasy. And in return [clears throat] 00:21:20.960 --> 00:21:23.357 >> But even Elon Musk was sort of like in agreement with that. 00:21:23.440 --> 00:21:25.797 >> Look, I I can't comment on what everyone's saying. I'm just telling you 00:21:25.880 --> 00:21:29.557 what's happening. I don't I don't know like I don't you know, we're down we 00:21:29.640 --> 00:21:34.867 have this massively growing business. I I barely get time for air. I'm It's like 00:21:34.950 --> 00:21:36.317 [laughter] we're building our products. 00:21:36.400 --> 00:21:40.757 We're we are going to own anthology in North fine-tuning the models, 00:21:40.840 --> 00:21:42.837 fine-tuning the open weight models, 00:21:42.920 --> 00:21:45.597 controlling the sovereign movement 00:21:45.680 --> 00:21:50.077 was primarily launched by us and Nvidia and like it's just like it's like it's 00:21:50.160 --> 00:21:53.957 like you're a rock band. I have these huge stadiums of people who want their 00:21:54.040 --> 00:21:57.117 sovereignty. So, I'm just telling people and I also 00:21:57.200 --> 00:22:01.797 >> high valuation on your stock. Do you think the market doesn't appreciate this? 00:22:01.880 --> 00:22:03.917 >> I I I look, I mean we are obviously so 00:22:04.000 --> 00:22:06.477 systemically undervalued. But it's okay. 00:22:06.560 --> 00:22:11.517 It's because people are Look, investors first of all, investors really did fall 00:22:11.600 --> 00:22:13.197 for I'm going to make a ton of money. 00:22:13.280 --> 00:22:14.877 There's no liability. It's all upside. 00:22:14.960 --> 00:22:18.757 All the alpha goes to them and therefore every bit of value of business is okay, 00:22:18.840 --> 00:22:21.757 great. Then these are very technical things. What does it mean to be 00:22:21.840 --> 00:22:25.837 sovereign? What does it mean to train a model? Are the Are the Like you have to 00:22:25.920 --> 00:22:29.157 The more you go into the granular details of our business and like how 00:22:29.240 --> 00:22:32.677 everybody wants to do it. The mere fact the sovereign movement is so much 00:22:32.760 --> 00:22:36.997 bigger. We need partners. We we cannot scale into our US market. I have people 00:22:37.080 --> 00:22:41.557 calling me from like very important allies and I'm like, "Look, we can't deploy." 00:22:41.640 --> 00:22:45.557 I can't take someone out of the US market. Like these are like crucial 00:22:45.640 --> 00:22:49.397 deployments where honestly we have the only thing that works. But it's like 00:22:49.480 --> 00:22:51.837 What are we going to do? >> Capacity? You just there's so much demand. 00:22:51.920 --> 00:22:56.037 >> It's like we we have we're mildly 4,500 00:22:56.120 --> 00:22:59.997 We're basically Think of it like every company. 3,600 people are the We're I 00:23:00.080 --> 00:23:05.397 would say are the best obviously the best in the world. Um and we're just we 00:23:05.480 --> 00:23:06.917 like every single one of them has like I 00:23:07.000 --> 00:23:12.477 have one guy from England. Like I literally to get the job done, I have 00:23:12.560 --> 00:23:15.277 like okay, just pretend you're three people work around the clock. I'll pay 00:23:15.360 --> 00:23:17.957 you for it. There's no other way to do it. 00:23:18.040 --> 00:23:19.637 >> Why? Cuz you can't find enough good people? 00:23:19.720 --> 00:23:22.877 >> Well, it's I mean it's like 00:23:22.960 --> 00:23:25.197 hiring and training people to be full 00:23:25.280 --> 00:23:27.997 Palantirians is is a is a process and 00:23:28.080 --> 00:23:30.197 the market is expanding way beyond the 00:23:30.280 --> 00:23:35.397 way what we can do. And so yeah, the market is expanding because 00:23:35.480 --> 00:23:40.317 like if you want to talk about customer demand, every single person I deal with 00:23:40.400 --> 00:23:45.397 is either on the sovereign thing or wants to be on it or even skeptics, you 00:23:45.480 --> 00:23:48.397 literally talk to them five seconds, you're like well, how do you feel about 00:23:48.480 --> 00:23:50.517 this and this? And they're like oh wow. 00:23:50.600 --> 00:23:54.517 It's like it's everywhere. And the depth of passion around this in 00:23:54.600 --> 00:23:58.397 enterprises like I've been at this for 20 years. I've been I was we we led the 00:23:58.480 --> 00:24:00.397 we led the anti-terror revolution, we 00:24:00.480 --> 00:24:02.597 led the the defense tech revolution, we 00:24:02.680 --> 00:24:05.637 led with Foundry like products that work 00:24:05.720 --> 00:24:10.077 revolution, especially in the US. We led the application layer, we're the first 00:24:10.160 --> 00:24:15.117 to do the sovereign thing and like and I've never seen anything like this. Like 00:24:15.200 --> 00:24:19.517 every single minute you're not in this country, you are losing money. 00:24:19.600 --> 00:24:23.557 >> Really quick, before we let you go, on on Dario as an example, what are the 00:24:23.640 --> 00:24:27.877 odds that that model, his model, doesn't survive? And what are the implications 00:24:27.960 --> 00:24:30.757 for stocks and everything else if it doesn't? 00:24:30.840 --> 00:24:34.597 >> I look, I don't like I I I I 00:24:34.680 --> 00:24:37.317 think he is a one of the great legends 00:24:37.400 --> 00:24:42.117 in tech and business and I'm not wishing them not to survive. I think obviously 00:24:42.200 --> 00:24:45.197 everyone's going to move to only using their product and application layer, 00:24:45.280 --> 00:24:49.477 that's good for them because and it's obviously better for 00:24:49.560 --> 00:24:52.037 us, but we were right. Um and 00:24:52.120 --> 00:24:54.437 the future is going to be enterprises 00:24:54.520 --> 00:24:57.597 using multiple models, some open, some 00:24:57.680 --> 00:25:01.677 closed. Prefer They're going to really where you have actual secrets and alpha, 00:25:01.760 --> 00:25:04.717 they're all going to move to open models. And where it's like marketing in 00:25:04.800 --> 00:25:08.557 Argentina, and the person wants to use a closed model, they'll do that. 00:25:08.640 --> 00:25:11.997 >> I just say I wonder about the politics of all of this. Bernie Sanders wants to 00:25:12.080 --> 00:25:16.797 now ban super intelligence. Elizabeth Warren also is calling for a pause 00:25:16.880 --> 00:25:18.797 >> It's not just the left. We have to be 00:25:18.880 --> 00:25:20.957 like, you know, I look I It's It's It's 00:25:21.040 --> 00:25:24.117 It the far right and far left. 00:25:24.200 --> 00:25:28.757 First of all, they're right about the unpopularity both of the movement and of 00:25:28.840 --> 00:25:33.077 the people. And they're taking advantage of the fact that the the the present 00:25:33.160 --> 00:25:34.837 presentation 00:25:34.920 --> 00:25:40.437 of the technology by the people doing it could not be worse. So, like my 00:25:40.520 --> 00:25:43.397 Honestly, if you want to take the far right and the far left, you can't hate 00:25:43.480 --> 00:25:45.157 the player because they understand the game. 00:25:45.240 --> 00:25:48.917 >> shock doc day. >> They They They They're just monetizing. 00:25:49.000 --> 00:25:51.037 They understand the political landscape. 00:25:51.120 --> 00:25:54.077 Are they right? No, but are there elements and truth in what they're 00:25:54.160 --> 00:25:57.517 saying? Of course. They're abusing elements of truth, but the core truth 00:25:57.600 --> 00:26:01.237 that they understand is you guys have wrecked yourself as if you 00:26:01.320 --> 00:26:03.357 do not have the sense God gave a goat. 00:26:03.440 --> 00:26:07.117 You are unlikable and you're going to be 50 times richer and we're not down for 00:26:07.200 --> 00:26:11.397 it. Then, the solutions are only going to help our adversaries. But again, it's 00:26:11.480 --> 00:26:14.837 not just Bernie Sanders and Liz Warren or whatever. These are people By the 00:26:14.920 --> 00:26:19.197 way, I happen to If you're listening, I agree with you guys on a lot of issues. 00:26:19.280 --> 00:26:22.917 Poor people in this country do not get a fair shake. 00:26:23.000 --> 00:26:25.717 That's true and I'm not certain that they're going to get that much of a 00:26:25.800 --> 00:26:28.797 better shake in 5 or 6 years. We agree. 00:26:28.880 --> 00:26:30.637 >> calling yourself a liberal progressive, Alex? 00:26:30.720 --> 00:26:32.637 >> I am an actual progressive. Actual 00:26:32.720 --> 00:26:37.877 progressives want poor people, working class people to to have better lives. 00:26:37.960 --> 00:26:41.437 What I disagree with them on is that this isn't going to help. But again, 00:26:41.520 --> 00:26:44.837 you can't go to the player and say, "You created the game. I'm not going to 00:26:44.920 --> 00:26:48.837 monetize it." They're absolutely doing what they need to do. 00:26:48.920 --> 00:26:51.277 We need to do a better job of doing what we need to do. 00:26:51.360 --> 00:26:54.757 >> Which is talking about how the benefits >> again and in any case thank you very 00:26:54.840 --> 00:26:57.757 much. >> Vampire Dracula here. If you found value 00:26:57.840 --> 00:27:00.637 in this video, please smash the like button. Let me know your favorite part 00:27:00.720 --> 00:27:03.717 of the video by commenting below. And most importantly, don't forget to 00:27:03.800 --> 00:27:09.920 subscribe and ding that bell. Until next time, keep it real.