Bill Gates: AI is the biggest technical thing ever in my lifetime (YouTube Video Transcript)

(00:00:00) Your YouTube transcript will appear here (00:00:00) Microsoft's co-founder Bill Gates (00:00:02) publishing a new essay on climate. It's (00:00:04) aimed at to deliver a message ahead of (00:00:06) COP 30 and he's suggesting that the (00:00:08) climate community should shift its (00:00:10) strategy from temperature and emission (00:00:12) targets to the impact on human lives. An (00:00:15) exclusive interview on that topic. I (00:00:16) also asked him if he is worried about (00:00:19) whether we are in an AI bubble. Here's (00:00:21) what he had to say. (00:00:24) We need to define bubble. If if what we (00:00:26) mean is like tulips in the Netherlands (00:00:29) that they eventually look back and said, (00:00:30) "What the heck? There was nothing there. (00:00:33) Those were just tulips." No, that's not (00:00:35) where we are. If you mean it's like the (00:00:38) internet bubble where in the end (00:00:41) something very profound happened. The (00:00:43) world was very different. Some companies (00:00:45) succeeded, but a lot of the companies (00:00:48) were kind of me too, fell behind, (00:00:52) uh burning capital companies. (00:00:54) Absolutely, (00:00:55) there are a ton of these investments um (00:00:59) that will be dead ends. And you look at (00:01:02) all of the investments that are being (00:01:03) made today by some of the big tech (00:01:05) companies. Some companies were not (00:01:06) making money on AI yet, but are making (00:01:08) massive commitments to chip makers and (00:01:11) data centers and say, "This math makes (00:01:13) sense." Um (00:01:15) the AI is the biggest (00:01:19) technical thing (00:01:21) ever in my lifetime. I mean, it is so (00:01:23) profound and therefore it's influence is (00:01:26) hard to overstate. And the economic (00:01:29) value, this is basically intelligence. (00:01:31) Uh you know, where you can get medical (00:01:33) advice, you can get a tutor, or you can (00:01:35) get somebody to help you design drugs. (00:01:38) So, the value is extremely high. Uh just (00:01:42) like creating the internet ended up (00:01:43) being in that very, very valuable. But (00:01:47) you have a frenzy and you know, some of (00:01:50) these companies will be glad they spent (00:01:52) all this money. Some of them, you know, (00:01:54) they'll commit to data centers that (00:01:56) whose electricity's is too expensive, (00:01:59) you know, that it could be done overseas (00:02:00) or they'll buy a generation of chips and (00:02:03) and, you know, they won't have captured (00:02:06) all their value before the next uh one (00:02:09) comes along. But you know, if you want (00:02:11) to be a tech company, you don't get to (00:02:13) say no, (00:02:15) uh (00:02:16) you know, let's check out of this race. (00:02:18) >> What do you think the public appetite (00:02:19) for this is ultimately going to be? And (00:02:21) I I ask because I think there's concerns (00:02:22) about two things that are happening (00:02:23) right now. One is energy costs. So, (00:02:26) there's a whole bunch of places where (00:02:27) people want to put data centers and uh (00:02:30) you know, the community says, "No, we (00:02:31) don't want this because of we don't want (00:02:32) to have to spend more money for power to (00:02:35) power our own homes." At the same time, (00:02:37) they're worried that if they actually (00:02:38) are successful, they'll also lose their (00:02:40) jobs. (00:02:41) We need to put things like TerraPower (00:02:44) nuclear reactors in places where it's (00:02:47) very clear that you're not raising the (00:02:51) the residents' electricity bills because (00:02:53) that's being put in there. Historically, (00:02:55) nuclear was done in a way that the (00:02:58) utility bore a lot of that liability. (00:03:01) That that business model won't get (00:03:04) repeated. So, we need to make sure (00:03:07) uh (00:03:07) to pick locations where the economics (00:03:10) and the political acceptance is very, (00:03:12) very strong. We have We don't have (00:03:14) permission to drive up people's (00:03:16) electricity's costs. Uh you know, in (00:03:19) terms of the jobs, this is going to take (00:03:22) some period of time. But yes, (00:03:25) uh (00:03:26) although it hasn't been seen in large (00:03:28) numbers, over the next several years, (00:03:30) there will be some impact on the job (00:03:32) market. Nowadays, when you say that, (00:03:34) some people are like, "Oh, how can you, (00:03:36) you know, (00:03:38) you know, say that? Isn't that going to (00:03:40) slow the US down in this race?" But it's (00:03:43) only honest for people (00:03:45) um (00:03:46) to speak frankly about the fact this (00:03:48) will have a big effect on the job (00:03:50) market. (00:03:51) And of course, uh Gates is a long-time (00:03:53) capitalist. So, I talked to him about (00:03:55) the industrial policy coming from the (00:03:57) White House, including the US taking (00:03:59) stakes in rare earth and chip companies. (00:04:01) Here's what he had to say. (00:04:04) The government (00:04:06) operates best when it's kind of (00:04:07) predictable. That is, when you know (00:04:09) what, you know, what are the tariffs (00:04:11) going to be for the next 20 years? Cuz (00:04:13) when you build a a factory, (00:04:15) uh so the government shouldn't be (00:04:16) changing its policy, you know, every (00:04:19) year or every day. Uh because, you know, (00:04:23) predictability is a very, very important (00:04:25) thing. Understanding (00:04:28) when the company's helping nascent (00:04:30) technology, (00:04:31) is it um doing that for the good of the (00:04:34) country and, you know, not and treating (00:04:37) companies equally, or will it want to (00:04:41) own part of those companies and not and (00:04:43) if somebody has better technology, (00:04:45) they'll favor the company they own part (00:04:47) of. You know, we need to understand what (00:04:49) those policies are. Yes, I think there's (00:04:51) a rush of people to say, "Wow, (00:04:54) uh you know, maybe that's how to, you (00:04:56) know, get to the front of the line for (00:04:58) government money." But the rules of the (00:05:00) game we're playing are pretty unclear (00:05:02) right now.