WEBVTT 00:00:00.080 --> 00:00:05.197 We need to make sure that we nail AI and we really have a fixed window over the 00:00:05.280 --> 00:00:08.437 next few years to make sure that America stays ahead. 00:00:08.520 --> 00:00:10.540 [music] 00:00:12.320 --> 00:00:14.637 >> Alex Wang is the 28-year-old founder and 00:00:14.720 --> 00:00:19.277 CEO of Scale AI, a data annotation platform that provides training data for 00:00:19.360 --> 00:00:22.637 machine learning models. You think we should prioritize 00:00:22.720 --> 00:00:26.957 leadership more than safety? Why do you feel that way? And why do you think that 00:00:27.040 --> 00:00:30.877 actually putting safety first could potentially endanger us? 00:00:30.960 --> 00:00:32.397 >> How does the Chinese Communist Party 00:00:32.480 --> 00:00:36.077 talk about AI? Going back to 2017, 2018, 00:00:36.160 --> 00:00:40.637 they enacted a kind of a whole of government approach to leading an AI, 00:00:40.720 --> 00:00:42.877 sort of an AI master plan, if you will. 00:00:42.960 --> 00:00:47.757 And what they talk about is they want to be the leader in AI by the year 2030. 00:00:47.840 --> 00:00:51.277 and they have an approach that integrates their economy, their 00:00:51.360 --> 00:00:55.597 government, and every part of the Chinese society to ensure that they're 00:00:55.680 --> 00:01:00.077 able to come out ahead. If you look at AI performance today, we always like to 00:01:00.160 --> 00:01:02.397 break down AI into its three core 00:01:02.480 --> 00:01:05.277 elements. Compute, algorithms, and data. 00:01:05.360 --> 00:01:07.997 On compute, America is still ahead. 00:01:08.080 --> 00:01:11.517 Algorithms were now basically neck andneck. DeepSeek basically proved that 00:01:11.600 --> 00:01:16.077 that they are just right on par with us on algorithms. And if you look at data, 00:01:16.160 --> 00:01:18.957 China is actually way ahead of the United States. And part of this is 00:01:19.040 --> 00:01:23.757 because they instituted all these programs to win on data over the past 00:01:23.840 --> 00:01:28.477 decade, starting with when they put together their uh global AI surveillance 00:01:28.560 --> 00:01:32.397 systems. You know, the United States needs to have a strategy for data 00:01:32.480 --> 00:01:36.077 dominance. And we need to make sure that we have a approach both in the public 00:01:36.160 --> 00:01:39.037 sector and the private sector to ensure that we win on data. 00:01:39.120 --> 00:01:41.997 >> What keeps you up at night? What is the worst case scenario? Is it going to be 00:01:42.080 --> 00:01:44.797 an extinction event like some people have warned? 00:01:44.880 --> 00:01:47.837 >> I think before we get to those kinds of 00:01:47.920 --> 00:01:52.557 scenarios of a human extinction event, there's there's very real things that 00:01:52.640 --> 00:01:56.637 I'm concerned about. One of the things I'm really concerned about is if China 00:01:56.720 --> 00:01:59.277 races ahead on AI and let's say they 00:01:59.360 --> 00:02:01.997 start applying AI to their military and 00:02:02.080 --> 00:02:06.717 their cyber operations more effectively than uh we do in the United States, then 00:02:06.800 --> 00:02:09.597 we're in for a really bad time. You know, they'll be able to hack into all 00:02:09.680 --> 00:02:12.957 of our systems. They'll be able to hack into our energy grid and shut it off. 00:02:13.040 --> 00:02:16.957 They'll be able to uh take all of our information. And who knows, if they 00:02:17.040 --> 00:02:21.197 really build an incredible advantage in AI, especially AI when it comes to 00:02:21.280 --> 00:02:25.997 military or defense technologies, they may use that to start conquering nearby 00:02:26.080 --> 00:02:28.957 countries. You know, we really don't know what's going to happen. 00:02:29.040 --> 00:02:30.397 >> And it's it's very interesting, too, cuz 00:02:30.480 --> 00:02:33.677 your parents are Chinese and you grew up in Los Alamos, and so you sort of have 00:02:33.760 --> 00:02:38.077 this deep understanding of security, [music] technology, 00:02:38.160 --> 00:02:41.277 and what all that means. So you have a very interesting perspective. 00:02:41.360 --> 00:02:43.837 >> It was very much so drilled into my 00:02:43.920 --> 00:02:45.837 childhood that that American leadership 00:02:45.920 --> 00:02:49.997 in science and technology is the core of 00:02:50.080 --> 00:02:54.077 why we as a country are able to maintain our national security, why we're able to 00:02:54.160 --> 00:02:58.957 maintain leadership. You know, it's so critical to ultimately how America stays 00:02:59.040 --> 00:03:03.037 ahead in the global order. >> So what are the jobs of the future that 00:03:03.120 --> 00:03:06.637 involve us working with AI and what does that look like for the average American? 00:03:06.720 --> 00:03:10.077 These are work opportunities that are going towards Americans and going 00:03:10.160 --> 00:03:12.557 towards people who uh who otherwise may 00:03:12.640 --> 00:03:17.437 not have exciting opportunities in a growing economic space like AI. And I 00:03:17.520 --> 00:03:21.357 think this is just going to keep accelerating. And I think jobs in AI are 00:03:21.440 --> 00:03:23.677 going to grow faster than anywhere else. 00:03:23.760 --> 00:03:26.077 You know, one of the other things that I definitely believe and I've been talking 00:03:26.160 --> 00:03:30.797 about is that we as humans are kind of going to become the managers of AI 00:03:30.880 --> 00:03:34.557 agents. It's almost like everybody in the world is going to get to be promoted 00:03:34.640 --> 00:03:36.477 to be a manager. But I do think that, 00:03:36.560 --> 00:03:37.917 you know, we're going to be able to 00:03:38.000 --> 00:03:41.597 utilize these AI agents to do more and 00:03:41.680 --> 00:03:45.237 have greater ambition on what each individual human is able to accomplish. 00:03:45.320 --> 00:03:47.437 [music] >> And it's decentralized, too. So, it's an 00:03:47.520 --> 00:03:51.277 interesting point you make about people in rural areas can participate. It 00:03:51.360 --> 00:03:54.317 doesn't have to just be in big cities like New York or San Francisco. [music] 00:03:54.400 --> 00:03:58.077 >> Yeah, exactly. That's one of I think the really exciting things is that because 00:03:58.160 --> 00:04:01.997 AI is a digital technology because all Americans are going to have access to 00:04:02.080 --> 00:04:04.237 technology, it really democratizes and 00:04:04.320 --> 00:04:09.117 and really uh you know spreads out the opportunity across the entire US. 00:04:09.200 --> 00:04:11.837 >> Is that what you would say excites you most or is there anything else that 00:04:11.920 --> 00:04:14.637 really excites you today about where you see AI going? 00:04:14.720 --> 00:04:16.477 >> The really exciting things are are just 00:04:16.560 --> 00:04:19.197 the degree to which it's going to enable 00:04:19.280 --> 00:04:21.677 people to accomplish way more. I mean, 00:04:21.760 --> 00:04:23.757 some of the things that that I find like 00:04:23.840 --> 00:04:28.717 at my at my core really exciting are how AI will be used in science and how it 00:04:28.800 --> 00:04:31.677 will be used to sort of accelerate scientific discoveries. 00:04:31.760 --> 00:04:35.117 >> Can you speak to scale? What's next for you? Obviously, you're busy working on 00:04:35.200 --> 00:04:38.717 issues like these, but but what what are what's the future of the company? 00:04:38.800 --> 00:04:41.757 >> We recently announced a program called Thunder Forge that we're working on 00:04:41.840 --> 00:04:44.717 actively with the Department of Defense to utilize [music] 00:04:44.800 --> 00:04:49.517 AI and AI agents for military planning and wargaming. We're very, very excited 00:04:49.600 --> 00:04:52.397 about that. I think there's so much more work to be done there. I think that's 00:04:52.480 --> 00:04:55.437 going to be something we work on [music] for decades to come. You know, it's not 00:04:55.520 --> 00:04:57.997 just like a a next year or next few year things 00:04:58.080 --> 00:05:00.237 >> because that's the future of warfare [snorts] it seems. 00:05:00.320 --> 00:05:02.157 >> Yes. Uh we we believe that 00:05:02.240 --> 00:05:03.837 fullheartedly. And then we're seeing a 00:05:03.920 --> 00:05:07.517 lot of exciting work with uh many uh 00:05:07.600 --> 00:05:11.757 many great American enterprises. We're seeing now that that American 00:05:11.840 --> 00:05:15.037 enterprises more so than than even enterprises globally are really 00:05:15.120 --> 00:05:19.117 embracing AI and want to figure out how they can utilize it to operate more 00:05:19.200 --> 00:05:23.037 efficiently, be more competitive, and deliver a greater result to their 00:05:23.120 --> 00:05:25.360 customers.