WEBVTT - Eric Schmidt Talks AI, TikTok

0:00:02.400 --> 0:00:07.280
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news is.

0:00:07.280 --> 0:00:10.800
<v Speaker 2>Your first have a AI expo for national competitiveness. So

0:00:11.280 --> 0:00:14.120
<v Speaker 2>where are we in US competitiveness when it comes to

0:00:14.120 --> 0:00:15.200
<v Speaker 2>out of show intelligence.

0:00:16.000 --> 0:00:17.920
<v Speaker 3>Well, the good news is the US is way ahead

0:00:17.920 --> 0:00:19.960
<v Speaker 3>of China and everybody else, and I think that's going

0:00:20.040 --> 0:00:23.720
<v Speaker 3>to continue for a while. To me, national competitiveness is

0:00:23.800 --> 0:00:26.600
<v Speaker 3>the challenge for the next ten or twenty years because

0:00:26.640 --> 0:00:30.960
<v Speaker 3>the Chinese are really focused on dominating certain industries and

0:00:31.360 --> 0:00:33.840
<v Speaker 3>we need to compete with them and make sure we win.

0:00:34.600 --> 0:00:37.640
<v Speaker 3>In the case of artificial intelligence, we are well ahead

0:00:38.040 --> 0:00:40.519
<v Speaker 3>two or three years. Probably we have China, which in

0:00:40.560 --> 0:00:43.280
<v Speaker 3>my world is in eternity. I think we're in pretty

0:00:43.280 --> 0:00:48.319
<v Speaker 3>good shape. And the crazy valuations, all the money, all

0:00:48.440 --> 0:00:52.240
<v Speaker 3>the new experimentation, this sort of enormous adoption of AI

0:00:52.360 --> 0:00:55.680
<v Speaker 3>that's occurring. We have a lot of implications on society

0:00:55.840 --> 0:00:57.400
<v Speaker 3>and very good for business.

0:00:57.640 --> 0:01:02.880
<v Speaker 2>What about well, ultimately, the of regulating artificial intelligence, is

0:01:02.920 --> 0:01:05.319
<v Speaker 2>there any risk that we stifle innovation and.

0:01:05.280 --> 0:01:05.960
<v Speaker 1>So doing.

0:01:07.440 --> 0:01:11.360
<v Speaker 3>There's always a risk of premature regulation. My simplest example

0:01:11.400 --> 0:01:14.679
<v Speaker 3>there is Europe. I've spent ten years trying to convince

0:01:14.720 --> 0:01:17.200
<v Speaker 3>Europe to actually innovate instead of regulate, and they just

0:01:17.319 --> 0:01:22.720
<v Speaker 3>keep regulating. The current act AI EU act is essentially regulation,

0:01:22.959 --> 0:01:25.920
<v Speaker 3>not investment in the future. So you can see that

0:01:25.959 --> 0:01:29.280
<v Speaker 3>Europe is highly unlikely to be relevant. China, of course,

0:01:29.360 --> 0:01:32.800
<v Speaker 3>is struggling because of chips shortages and so forth. All

0:01:32.840 --> 0:01:35.200
<v Speaker 3>that they're ready to win if they can get the

0:01:35.240 --> 0:01:38.000
<v Speaker 3>harder that they need, and the rest of the world.

0:01:37.760 --> 0:01:39.120
<v Speaker 1>Is not focused enough on this.

0:01:39.720 --> 0:01:43.280
<v Speaker 3>So the good news from myself my position in terms

0:01:43.280 --> 0:01:46.360
<v Speaker 3>of America here we are in DC for national competitiveness.

0:01:46.640 --> 0:01:48.880
<v Speaker 1>We're the likely winner if we don't screw it up.

0:01:49.840 --> 0:01:50.880
<v Speaker 1>What could screw it up?

0:01:51.200 --> 0:01:54.160
<v Speaker 2>Could it be a lack of government focus? How much

0:01:54.320 --> 0:01:56.520
<v Speaker 2>you how would you grade the current administration when it

0:01:56.520 --> 0:01:58.760
<v Speaker 2>comes to a focus on innovation and funding it.

0:02:00.160 --> 0:02:02.520
<v Speaker 3>We've been working with the US government and they've taken

0:02:02.520 --> 0:02:05.840
<v Speaker 3>a largely handoff, hands off, you know, tell us what's

0:02:05.880 --> 0:02:09.040
<v Speaker 3>going on approach, which has worked pretty well. There are

0:02:09.040 --> 0:02:11.800
<v Speaker 3>now voluntary commitments from all the companies. That's a good

0:02:11.880 --> 0:02:16.040
<v Speaker 3>first step. I'm sure there will be areas of real problems,

0:02:16.080 --> 0:02:19.720
<v Speaker 3>you know, cyber dangers, biological dangers and so forth, and

0:02:19.800 --> 0:02:21.760
<v Speaker 3>I would expect in the next ten years you'll see

0:02:22.040 --> 0:02:26.000
<v Speaker 3>real regulations in some of these spaces because of potential harm. Today,

0:02:26.080 --> 0:02:29.160
<v Speaker 3>it's unlikely you're going to have serious harm unless you

0:02:29.200 --> 0:02:31.160
<v Speaker 3>think that the deep fakes are a serious harm, which

0:02:31.200 --> 0:02:32.560
<v Speaker 3>they may be in some cases.

0:02:33.560 --> 0:02:34.840
<v Speaker 1>So we're okay at the moment.

0:02:35.360 --> 0:02:38.840
<v Speaker 2>Would you worry about the next administration if it was

0:02:38.960 --> 0:02:41.320
<v Speaker 2>indeed Underbiden, if it was a Trump administration.

0:02:44.000 --> 0:02:47.239
<v Speaker 3>We are steadfastly bipartisan on this issue because the development

0:02:47.280 --> 0:02:50.280
<v Speaker 3>of AI benefits everybody in the nation. It benefits in

0:02:50.360 --> 0:02:53.480
<v Speaker 3>terms of, you know, curing cancer, helping with climate change,

0:02:53.880 --> 0:02:58.720
<v Speaker 3>new materials, making energy more efficient, distribution use, and so forth,

0:02:58.760 --> 0:03:03.359
<v Speaker 3>and so on. Actional competitiveness should be a bipartisan issue

0:03:03.600 --> 0:03:05.840
<v Speaker 3>in each administration. You have to get to know them,

0:03:05.880 --> 0:03:08.440
<v Speaker 3>you have to get them up to speed. One of

0:03:08.440 --> 0:03:11.720
<v Speaker 3>the strange things about America is that we change over

0:03:11.760 --> 0:03:14.639
<v Speaker 3>the entire government every four or eight years, and then

0:03:14.680 --> 0:03:17.040
<v Speaker 3>you have this whole new bunch of people who have

0:03:17.080 --> 0:03:19.720
<v Speaker 3>to learn both how the government works and also they

0:03:19.760 --> 0:03:23.280
<v Speaker 3>come in with an agenda. I'm not too worried about

0:03:23.360 --> 0:03:27.679
<v Speaker 3>it because everybody benefits from this, whether you're a Republican

0:03:27.760 --> 0:03:31.679
<v Speaker 3>or a Democrat, whether you're a liberal or a conservatives.

0:03:30.560 --> 0:03:33.320
<v Speaker 2>Nuanced and it's also a nuanced relationship that corporates have

0:03:33.360 --> 0:03:36.480
<v Speaker 2>to work with the United States and indeed with China

0:03:36.520 --> 0:03:38.520
<v Speaker 2>at the moment. And I'm really interested you say, perhaps

0:03:38.600 --> 0:03:40.680
<v Speaker 2>China is being pushed behind because of its lack of

0:03:40.720 --> 0:03:43.880
<v Speaker 2>access to chips all we currently navigating that nuance.

0:03:43.920 --> 0:03:48.960
<v Speaker 3>Well, Well, the Trump administration followed by the Biden administration

0:03:49.040 --> 0:03:51.680
<v Speaker 3>did a good job in getting those rules in place.

0:03:52.560 --> 0:03:55.520
<v Speaker 3>It's a good example and not a common example, of

0:03:55.520 --> 0:03:59.800
<v Speaker 3>how a targeted intervention has a positive measured outcome from

0:03:59.840 --> 0:04:01.920
<v Speaker 3>the standpoint of the national security.

0:04:02.760 --> 0:04:07.480
<v Speaker 1>So I think we're good there. It's very important.

0:04:07.000 --> 0:04:10.680
<v Speaker 3>And I led the National Security Commission on Artificial Intelligence

0:04:11.520 --> 0:04:14.440
<v Speaker 3>a few years ago for the Congress. It's very important

0:04:14.840 --> 0:04:17.400
<v Speaker 3>that we get more basic money out of the government

0:04:17.440 --> 0:04:18.240
<v Speaker 3>for basic research.

0:04:18.320 --> 0:04:19.680
<v Speaker 1>We're building intelligence.

0:04:19.720 --> 0:04:22.719
<v Speaker 3>We're building a new form of non human intelligence that

0:04:22.760 --> 0:04:26.440
<v Speaker 3>will change the world forever in a good way. And furthermore,

0:04:26.480 --> 0:04:28.640
<v Speaker 3>it needs to be done with our partner countries. These

0:04:28.640 --> 0:04:31.400
<v Speaker 3>are the five Eyes of the UK, that sort of

0:04:31.400 --> 0:04:34.159
<v Speaker 3>thing some of the European countries. And also it needs

0:04:34.200 --> 0:04:38.160
<v Speaker 3>to be done with American values. So using China as

0:04:38.160 --> 0:04:41.200
<v Speaker 3>the bogeyman, imagine if they were in charge of the Internet,

0:04:41.240 --> 0:04:43.120
<v Speaker 3>and they were in charge of the rules. And you

0:04:43.160 --> 0:04:46.000
<v Speaker 3>imagine how very different your experience as a user on

0:04:46.040 --> 0:04:48.599
<v Speaker 3>the Internet would be today. You can show up, you

0:04:48.640 --> 0:04:51.200
<v Speaker 3>can be anonymous, you can do whatever you want within reason.

0:04:51.880 --> 0:04:54.240
<v Speaker 3>That kind of freedom is central to the spirit of

0:04:54.240 --> 0:04:56.159
<v Speaker 3>the Internet, and it's important to be preserved.

0:04:56.800 --> 0:05:00.560
<v Speaker 2>Talking about a bogeyman and a Chinese related bogaman, TikTok

0:05:00.640 --> 0:05:04.320
<v Speaker 2>became the bogaman that many are now trying to regulate,

0:05:04.360 --> 0:05:07.680
<v Speaker 2>to indeed ban, or to sell. There's been some reporting

0:05:07.720 --> 0:05:14.240
<v Speaker 2>that you've been interested, alongside Manution in potentially purchasing us TikTok.

0:05:14.279 --> 0:05:15.040
<v Speaker 1>Would you still be.

0:05:17.600 --> 0:05:19.920
<v Speaker 3>I'm not currently looking at that. I looked at it

0:05:19.920 --> 0:05:23.520
<v Speaker 3>for a while. My personal view on this is you're

0:05:23.520 --> 0:05:29.159
<v Speaker 3>better off regulating than banning or a judicial action. All

0:05:29.279 --> 0:05:31.520
<v Speaker 3>the big tech companies are now in the hands of

0:05:31.520 --> 0:05:34.560
<v Speaker 3>the DOJ in various legal fights. I would prefer to

0:05:34.560 --> 0:05:36.880
<v Speaker 3>see a regulatory regime that is sort of has the

0:05:36.960 --> 0:05:39.880
<v Speaker 3>right incentives and the right prohibitions for all of these things.

0:05:40.240 --> 0:05:42.680
<v Speaker 3>My own view of TikTok is that TikTok is not

0:05:42.720 --> 0:05:45.560
<v Speaker 3>really social media, it's really television, and that you can

0:05:45.640 --> 0:05:48.320
<v Speaker 3>regulate television by the equivalent of the equal time rule,

0:05:48.400 --> 0:05:50.039
<v Speaker 3>but somehow we're not having that conversation.

0:05:51.200 --> 0:05:53.960
<v Speaker 2>What's interesting is, of course, you say, many big tech

0:05:54.000 --> 0:05:57.640
<v Speaker 2>companies are currently well entangled from a time perspective with

0:05:57.680 --> 0:06:00.359
<v Speaker 2>the DOJ one of them being where we know you

0:06:00.400 --> 0:06:03.560
<v Speaker 2>holm from? Of course the previous CEO of Google. Where

0:06:03.600 --> 0:06:06.359
<v Speaker 2>did you rate from a personal perspective? Where Google is

0:06:06.360 --> 0:06:07.880
<v Speaker 2>in the AI competition race.

0:06:09.160 --> 0:06:12.400
<v Speaker 3>Well, it's interesting that almost all of the technology you're

0:06:12.400 --> 0:06:16.720
<v Speaker 3>seeing today was invented at Google about five to ten

0:06:16.800 --> 0:06:20.360
<v Speaker 3>years ago. And it's one of the sort of great

0:06:20.400 --> 0:06:23.320
<v Speaker 3>things that during that period Google had a roughly in

0:06:23.360 --> 0:06:26.120
<v Speaker 3>my estimate, two thirds of the world's talent in AI

0:06:26.279 --> 0:06:28.320
<v Speaker 3>was working at so Google had a very very strong

0:06:28.360 --> 0:06:31.920
<v Speaker 3>head start. I think Google is a complicated place, is

0:06:31.960 --> 0:06:35.800
<v Speaker 3>a lot going on. Open Ai, in particular came out

0:06:35.800 --> 0:06:38.440
<v Speaker 3>with GPT three three point five and four and did

0:06:38.480 --> 0:06:42.200
<v Speaker 3>a fantastic job. I think that was the competitive nudge

0:06:42.520 --> 0:06:45.160
<v Speaker 3>that has gotten Google back in the game. What I

0:06:45.279 --> 0:06:48.920
<v Speaker 3>like now is you have these two huge companies, Microsoft

0:06:48.960 --> 0:06:53.440
<v Speaker 3>and open Ai together and Google Google Alphabet again operating together,

0:06:53.880 --> 0:06:57.279
<v Speaker 3>and you have them putting billions of dollars hardware and

0:06:57.680 --> 0:07:01.280
<v Speaker 3>enormous software teams to invent this new future. This is

0:07:01.320 --> 0:07:05.920
<v Speaker 3>not to take away from Anthropic, another competitor. Obviously, whatever

0:07:05.960 --> 0:07:08.159
<v Speaker 3>Elon is doing with x dot ai, he's raising a

0:07:08.200 --> 0:07:11.040
<v Speaker 3>huge amount of money, and Meta has recently released a

0:07:11.080 --> 0:07:14.360
<v Speaker 3>four hundred billion parameter modelers and the process of releading

0:07:14.360 --> 0:07:18.120
<v Speaker 3>it that looks really, really good. So that competition is

0:07:18.160 --> 0:07:22.000
<v Speaker 3>the right answer to the regulators and to everyone's question.

0:07:22.560 --> 0:07:28.480
<v Speaker 3>Competition brings out enormous value to consumers. And eventually people

0:07:28.520 --> 0:07:30.800
<v Speaker 3>will see this and they'll say, oh my god, look

0:07:30.840 --> 0:07:33.000
<v Speaker 3>at what this does. And it's free. And of course

0:07:33.080 --> 0:07:36.120
<v Speaker 3>all that money was paid out of the markets. Billions

0:07:36.120 --> 0:07:38.440
<v Speaker 3>of dollars that's going in right now, and they'll monetize

0:07:38.440 --> 0:07:39.080
<v Speaker 3>it somehow.

0:07:39.640 --> 0:07:40.440
<v Speaker 1>Does it have to.

0:07:40.400 --> 0:07:44.840
<v Speaker 2>Be led by the big corporates? How much well optimism

0:07:44.880 --> 0:07:48.160
<v Speaker 2>do you have that the startup culture is what's going

0:07:48.200 --> 0:07:50.360
<v Speaker 2>to breed the innovation as well you as an active

0:07:50.360 --> 0:07:52.680
<v Speaker 2>investor as well as someone who's trying to breed competitiveness

0:07:52.720 --> 0:07:53.880
<v Speaker 2>from a national perspective.

0:07:55.520 --> 0:07:57.600
<v Speaker 3>I've tried to invest in most of the companies because

0:07:57.640 --> 0:07:59.400
<v Speaker 3>I frankly can't figure out which ones are going to

0:07:59.400 --> 0:08:03.560
<v Speaker 3>be successful. One of the most interesting startup successes is Anthropic,

0:08:03.640 --> 0:08:06.160
<v Speaker 3>which is now valued very high, started off as a

0:08:06.160 --> 0:08:08.920
<v Speaker 3>bunch of founders who came out of open Ai, who

0:08:09.000 --> 0:08:11.320
<v Speaker 3>had invented some of the technology but not all of it,

0:08:11.360 --> 0:08:13.960
<v Speaker 3>and they built an unusually good product. Their big partner

0:08:14.400 --> 0:08:17.600
<v Speaker 3>is Amazon, and that again makes sense to me, right

0:08:17.640 --> 0:08:19.800
<v Speaker 3>because they can do it faster than Amazon. They're very,

0:08:19.920 --> 0:08:23.200
<v Speaker 3>very focused. So this new industrial structure of these deep

0:08:23.200 --> 0:08:25.240
<v Speaker 3>partnerships is not a bad thing.

0:08:25.680 --> 0:08:25.920
<v Speaker 1>Look.

0:08:25.960 --> 0:08:29.520
<v Speaker 3>The important thing is the arrival of these large language

0:08:29.520 --> 0:08:32.280
<v Speaker 3>models is a new economic problem in our industry because

0:08:32.320 --> 0:08:36.240
<v Speaker 3>the amount of capital. In my fifty years of doing software,

0:08:36.960 --> 0:08:39.880
<v Speaker 3>we never had capital constraints. We just had software people

0:08:40.080 --> 0:08:42.480
<v Speaker 3>and the software. We always had whatever capital we need.

0:08:42.840 --> 0:08:45.640
<v Speaker 3>Now in order to do anything, you need ten thousand

0:08:45.760 --> 0:08:49.439
<v Speaker 3>or twenty thousand in video H one hundreds B two hundreds,

0:08:49.679 --> 0:08:51.480
<v Speaker 3>which cost forty thousand dollars a pop.

0:08:52.120 --> 0:08:53.079
<v Speaker 1>So you do the math.

0:08:53.480 --> 0:08:56.480
<v Speaker 3>All of a sudden, your idea to get started has

0:08:56.520 --> 0:09:01.240
<v Speaker 3>to have a billion dollar valuation. That's event in a

0:09:01.280 --> 0:09:05.000
<v Speaker 3>software industry, and an unfortunately one which I think tends

0:09:05.040 --> 0:09:08.880
<v Speaker 3>to reward large companies as a bush of the smaller companies.

0:09:09.200 --> 0:09:12.680
<v Speaker 3>The best scenarios where the large companies are investing heavily,

0:09:12.800 --> 0:09:15.520
<v Speaker 3>and the small companies are right up trying to invent

0:09:15.600 --> 0:09:17.679
<v Speaker 3>the new future. And every once in a while one

0:09:17.679 --> 0:09:20.840
<v Speaker 3>of those small companies becomes some huge monolith. Look at Meta,

0:09:21.080 --> 0:09:23.600
<v Speaker 3>look at TikTok. You get the idea and.

0:09:23.600 --> 0:09:26.640
<v Speaker 2>You mentioned, well a global company there with TikTok. What

0:09:26.720 --> 0:09:29.439
<v Speaker 2>about European startups? Could we have a giant out of

0:09:29.480 --> 0:09:31.840
<v Speaker 2>Europe or is regulation just ended up? Would you invest

0:09:31.840 --> 0:09:32.960
<v Speaker 2>in a European startup?

0:09:34.040 --> 0:09:34.640
<v Speaker 1>I have been.

0:09:34.520 --> 0:09:38.480
<v Speaker 3>Investing in France personally because the French ecosystem, in particular

0:09:38.559 --> 0:09:41.760
<v Speaker 3>under President Macron, is much more silk and valley like.

0:09:43.040 --> 0:09:45.920
<v Speaker 3>It's going to be rare to see the successes in

0:09:45.960 --> 0:09:49.880
<v Speaker 3>Europe that you see in America in the UK, and

0:09:49.920 --> 0:09:52.800
<v Speaker 3>the fundamental reason is they can't get the capital and

0:09:52.800 --> 0:09:57.120
<v Speaker 3>the regulatory structure is so prohibitive. It's really sad because

0:09:57.120 --> 0:10:01.520
<v Speaker 3>the European talent pool is amazing. The opportunity to build

0:10:01.600 --> 0:10:04.640
<v Speaker 3>companies in terms of impact is huge, and I think

0:10:04.679 --> 0:10:07.280
<v Speaker 3>frankly the government and the regulation is slowing them down.

0:10:07.720 --> 0:10:09.320
<v Speaker 1>So I'm not writing them off.

0:10:09.360 --> 0:10:12.520
<v Speaker 3>I'm doing some investing, but so far the action will

0:10:12.520 --> 0:10:15.400
<v Speaker 3>be in the US, the UK and Canada, where the

0:10:15.440 --> 0:10:19.320
<v Speaker 3>regulatory structures are much more permissible to economic investment.

0:10:20.240 --> 0:10:23.400
<v Speaker 2>And when we started this conversation, you said, the US

0:10:23.480 --> 0:10:25.199
<v Speaker 2>is leading as long as they don't screw it up.

0:10:25.640 --> 0:10:28.120
<v Speaker 2>One piece of advice to ensure that the US doesn't.

0:10:32.040 --> 0:10:35.400
<v Speaker 3>Let's start by saying, the US capital system is extraordinary.

0:10:36.040 --> 0:10:39.600
<v Speaker 3>You can raise billions of dollars based on an idea

0:10:39.720 --> 0:10:41.640
<v Speaker 3>and a promise of a scale business.

0:10:42.040 --> 0:10:43.240
<v Speaker 1>Let's not screw that up.

0:10:43.720 --> 0:10:47.720
<v Speaker 3>The second thing is, let's not regulate these companies before

0:10:47.760 --> 0:10:51.040
<v Speaker 3>they do something. Give them a chance to figure out

0:10:51.320 --> 0:10:55.360
<v Speaker 3>how to monetize these spaces and build new businesses. Look

0:10:55.400 --> 0:10:58.960
<v Speaker 3>at the tech companies that define the US stock market.

0:10:59.320 --> 0:11:02.520
<v Speaker 1>Look at the value created. For the last couple of years, the.

0:11:02.679 --> 0:11:05.680
<v Speaker 3>Sort of Lucky seven, the Famous four, whatever you want

0:11:05.679 --> 0:11:08.959
<v Speaker 3>to call them, have driven the US stock market as

0:11:09.000 --> 0:11:12.280
<v Speaker 3>the primary drivers of wealth for the whole nation. Right,

0:11:12.679 --> 0:11:14.840
<v Speaker 3>So the game that we're in is one where we

0:11:14.880 --> 0:11:17.600
<v Speaker 3>want to create another one, and another one and another one.

0:11:17.960 --> 0:11:22.280
<v Speaker 3>That opportunity is how economics works in America, and we're

0:11:22.280 --> 0:11:25.600
<v Speaker 3>in a position to provide global platforms at scale using

0:11:25.640 --> 0:11:29.960
<v Speaker 3>the combination of American research, American policy, and all of

0:11:29.960 --> 0:11:33.480
<v Speaker 3>this money that's raisable This is a godsend. Every other

0:11:33.480 --> 0:11:36.200
<v Speaker 3>country in the world would want this. Let's keep it going.