WEBVTT - boomers, doomers, and the new empire 0:00:11.280 --> 0:00:14.520 Empire of Ai. You're using the word empire here. Yeah, 0:00:14.600 --> 0:00:16.480 Can you tell me a little bit more about that decision? 0:00:16.720 --> 0:00:21.920 Yeah, So the term empire in the title is specifically 0:00:21.960 --> 0:00:23.960 an argument that I make in the book that we 0:00:24.079 --> 0:00:26.560 really need to start thinking of companies like open Ai 0:00:26.680 --> 0:00:28.320 as new forms of empire. 0:00:29.800 --> 0:00:32.199 Karen Howe is a journalist and the author of a 0:00:32.240 --> 0:00:35.920 new book called Empire of Ai Dreams and Nightmares and 0:00:36.000 --> 0:00:39.560 Sam Altman's Open Ai, which just came out last month. 0:00:39.880 --> 0:00:42.320 In case you're not familiar with her work, she herself 0:00:42.400 --> 0:00:45.160 is a data engineer, and she's also the first person 0:00:45.200 --> 0:00:48.240 to ever really cover open Ai, the company that makes 0:00:48.320 --> 0:00:51.240 chat gpt. On the surface, the book is a story 0:00:51.240 --> 0:00:53.760 of the rise of open Ai and the story of 0:00:53.800 --> 0:00:57.000 its founder, Sam Altman, and if you like drama, let 0:00:57.000 --> 0:00:59.720 me tell you there's a lot of it. Someone could 0:00:59.800 --> 0:01:02.680 definitely make a Netflix series about this. And they are 0:01:02.720 --> 0:01:06.600 also comparisons to be made between Sam Altman and Steve Jobs, 0:01:07.040 --> 0:01:09.960 and we will get to that. But before we do anything, 0:01:10.360 --> 0:01:12.280 we should talk about a word that a lot of 0:01:12.280 --> 0:01:15.240 the reviews I've seen are kind of glossing over, which 0:01:15.280 --> 0:01:20.160 is weird because it's the first word in the title empire, and. 0:01:20.120 --> 0:01:23.119 The reason is four different features that empires of AI 0:01:23.319 --> 0:01:26.520 share with empires of old. The first one is that 0:01:26.560 --> 0:01:28.840 they lay claim to resources that are not their own, 0:01:28.959 --> 0:01:32.039 but they reinterpret the rules to suggest that those resources 0:01:32.040 --> 0:01:35.119 were always their own. That refers to how these companies 0:01:35.280 --> 0:01:37.920 just scrape the data on the Internet, same with the 0:01:37.920 --> 0:01:41.120 intellectual property that these companies just take, like the work 0:01:41.160 --> 0:01:43.319 of artists, the work of writers, the work of creators. 0:01:43.440 --> 0:01:46.520 The second feature of empires is that they engage in 0:01:46.520 --> 0:01:48.960 a lot of labor exploitation, and that refers not just 0:01:49.000 --> 0:01:52.520 to the fact that these companies they contract a lot 0:01:52.560 --> 0:01:55.320 of workers, often in the Global South or other economically 0:01:55.400 --> 0:01:59.320 vulnerable communities in the Global North, and pay them extraordinarily 0:01:59.360 --> 0:02:04.840 small amounts of money to do extremely exploitative work data annotation, 0:02:04.960 --> 0:02:09.880 data cleaning, data preparation, content moderation, where workers are left 0:02:09.960 --> 0:02:12.680 with the same level of trauma that social media content 0:02:12.720 --> 0:02:15.400 moderators were left with. It also refers to the fact 0:02:15.480 --> 0:02:21.080 that these companies are inherently building labor automating technologies. So 0:02:21.200 --> 0:02:25.680 opening eyes definition of artificial general intelligence is highly autonomous 0:02:25.720 --> 0:02:31.360 systems that outperform humans at most economically valuable work and 0:02:31.440 --> 0:02:35.480 so they're explicitly saying we're going to automate away the 0:02:35.480 --> 0:02:38.440 things that people usually get paid for, and that in 0:02:38.480 --> 0:02:40.919 and of itself is a labor exploitation. So there's labor 0:02:40.919 --> 0:02:43.160 exploitation happening on the way in and happening on the 0:02:43.160 --> 0:02:43.800 way out. 0:02:44.080 --> 0:02:47.760 Karen is not kidding. In the book, she travels all 0:02:47.800 --> 0:02:50.600 over the world and speaks to people who work in 0:02:50.680 --> 0:02:54.040 all levels of the AI industry. And what you've found 0:02:54.280 --> 0:02:57.080 was that countries that were undergoing some kind of economic 0:02:57.160 --> 0:03:00.079 crisis were being targeted by the industry as places to 0:03:00.200 --> 0:03:03.639 hire workers who train AI models to understand what kind 0:03:03.680 --> 0:03:07.240 of stuff to allow and what to block. This is 0:03:07.320 --> 0:03:11.040 work that only a human can do. So who's this 0:03:11.120 --> 0:03:14.960 empire for? And where do you and I fit in here? 0:03:21.360 --> 0:03:31.120 I'm afraid Kaleidoscope and iHeart podcasts. This is kill switch. 0:03:32.040 --> 0:03:33.079 I'm doctor Thomas. 0:03:34.000 --> 0:04:00.160 I'm sorry, I'm goodbye. 0:04:03.160 --> 0:04:06.920 Getting back to Karen's definitions of colonial empires, the first 0:04:07.040 --> 0:04:09.400 is that they change the rules so that when they 0:04:09.440 --> 0:04:12.640 find something like oil in the ground or art and 0:04:12.760 --> 0:04:15.560 literature on the internet, they can say that it belongs 0:04:15.560 --> 0:04:18.320 to them. The second is that they exploit people for 0:04:18.360 --> 0:04:21.600 their labor. The third is that they control what people 0:04:21.640 --> 0:04:24.640 can know about them. In the case of the AI industry, 0:04:24.920 --> 0:04:28.160 this can look like companies just finding the top researchers 0:04:28.160 --> 0:04:31.360 who are working at universities, paying them lots of money 0:04:31.400 --> 0:04:33.839 to drop their university job and come to the company, 0:04:34.440 --> 0:04:37.839 and then censoring anything that those researchers write that is 0:04:37.880 --> 0:04:42.440 critical of their technology or of how they treat the environment. Again, 0:04:42.720 --> 0:04:46.080 controlling what the public knows about them. But I think 0:04:46.080 --> 0:04:49.400 it's the fourth element where things start to really get interesting. 0:04:49.800 --> 0:04:52.200 And then the last future is that empires always have 0:04:52.279 --> 0:04:56.600 this narrative that there are good empires and their evil empires. 0:04:57.040 --> 0:05:00.599 So they the good empire need to be empire in 0:05:00.600 --> 0:05:02.880 the first place and engage in all this resource extraction 0:05:02.920 --> 0:05:06.160 and labor exploitation because that is what will make them 0:05:06.200 --> 0:05:08.760 strong enough to beat back the evil empire. So back 0:05:08.760 --> 0:05:11.520 in the day and during European colonialism, the British Empire 0:05:11.520 --> 0:05:13.240 would say that they were better than the Dutch Empire. 0:05:13.279 --> 0:05:14.919 The French Empire would say they were better than the 0:05:14.920 --> 0:05:19.640 British Empire. And part of being the good empire is 0:05:19.680 --> 0:05:23.520 that what they were ultimately doing was civilizing the world. 0:05:24.160 --> 0:05:28.200 They were bringing progress and modernity to everyone, and they 0:05:28.560 --> 0:05:33.120 were giving humanity the chance to enter heaven instead of hell. 0:05:33.480 --> 0:05:39.520 And this is literally the language that AI people use 0:05:39.600 --> 0:05:42.320 these days. They talk about heaven and hell. They talk 0:05:42.360 --> 0:05:48.320 about recreating God and about bringing the next era of 0:05:48.360 --> 0:05:50.559 civilization into being. 0:05:51.440 --> 0:05:54.320 Yeah, and I mean Sam Altman twenty twenty three going 0:05:54.360 --> 0:05:58.640 before Congress and saying, look, if we don't put a 0:05:58.640 --> 0:06:01.440 whole bunch of resources in AA, guess who will China? 0:06:01.800 --> 0:06:04.440 Would you like China to win? I don't think you do. 0:06:04.760 --> 0:06:07.320 I know that you're scared of China. Let's play nice 0:06:07.360 --> 0:06:10.680 here so that you know, play nice specifically with our industry. 0:06:10.720 --> 0:06:13.920 Also play nice with my company so that we can 0:06:13.960 --> 0:06:16.919 come out on top, which again is very similar to 0:06:17.320 --> 0:06:22.320 how an empire justifies itself, which is, if we don't 0:06:22.360 --> 0:06:25.280 do this, the bad guys will, such as you lay 0:06:25.360 --> 0:06:29.000 on the book open AI really starts out by saying, listen, 0:06:29.640 --> 0:06:32.680 AI is going to happen. AJI is going to happen. 0:06:33.120 --> 0:06:36.039 It needs to be the good guys. We're the good guys. 0:06:36.080 --> 0:06:37.920 And I think that's something that you do in the 0:06:37.960 --> 0:06:40.920 book is I think we've gotten really used to this 0:06:41.080 --> 0:06:44.839 idea that everything was inevitable this all was going to happen, 0:06:44.960 --> 0:06:47.200 and the stuff that happens tomorrow it was always going 0:06:47.279 --> 0:06:49.719 to happen, And the stuff that happens three years from now, 0:06:49.760 --> 0:06:52.720 it was always going to happen. But you put it 0:06:52.760 --> 0:06:56.320 together that none of this is actually inevitable, and not 0:06:56.400 --> 0:06:59.760 only that, it's decisions are being made, and actually a 0:06:59.760 --> 0:07:03.320 lot of these decisions are being made by a pretty 0:07:03.360 --> 0:07:06.159 small handful of people in Silicon Valley. 0:07:06.520 --> 0:07:09.040 Absolutely, this is definitely one of the core messages that 0:07:09.080 --> 0:07:11.600 I hope people can take away from the book is 0:07:12.120 --> 0:07:14.920 technology is very much a product of human choices, and 0:07:15.000 --> 0:07:18.400 it just so happens that AI has been the product 0:07:18.480 --> 0:07:20.960 of a very small handful of people's choices who have 0:07:21.120 --> 0:07:25.120 a very what I would argue, narrow world view, a 0:07:25.240 --> 0:07:27.920 narrow view of how the world works now and how 0:07:28.000 --> 0:07:31.880 it should continue to work. And if I had to 0:07:31.880 --> 0:07:34.960 point to one decision that was not inevitable, but that 0:07:35.040 --> 0:07:37.640 Opening Eye sort of ushered in with the shape of 0:07:37.680 --> 0:07:41.160 AI today, it is the fact that they decided to 0:07:41.360 --> 0:07:47.480 scale this technology aggressively, and that happened because open Ai 0:07:47.680 --> 0:07:51.640 early on identified that they had to be number one 0:07:51.760 --> 0:07:53.960 based on this. There's the evil Empire out there, and 0:07:53.960 --> 0:07:58.800 we the Good Empire, have to race aggressively. And they 0:07:58.880 --> 0:08:03.240 realized that the fact fastest and easiest way to get 0:08:03.280 --> 0:08:06.360 to number one was by taking existing techniques within the 0:08:06.360 --> 0:08:10.040 AI research field and blowing it up with an unprecedented 0:08:10.040 --> 0:08:14.120 amount of data and an unprecedented amount of computational resources. 0:08:14.120 --> 0:08:16.920 So they realized, if we can build the largest supercomputers 0:08:16.920 --> 0:08:21.200 in the world, we will have the best chance at 0:08:21.360 --> 0:08:25.440 dominating this technology. All of a sudden, all of the 0:08:25.480 --> 0:08:28.600 tech companies saw what opening I did and when we 0:08:28.640 --> 0:08:31.280 want to do that too, And suddenly you see a 0:08:31.360 --> 0:08:34.679 step change in the sheer amount of resources that are 0:08:34.679 --> 0:08:37.720 now going into developing this technology, and it very much 0:08:37.760 --> 0:08:42.600 becomes a scale at all costs paradigm for AI development. 0:08:42.960 --> 0:08:48.480 This very specific scaling decision is what leads to significant 0:08:48.480 --> 0:08:51.440 amounts of labor exploitation and is what leads to significant 0:08:51.480 --> 0:08:55.320 amounts of environmental harm, because that is when you start 0:08:55.360 --> 0:08:58.559 talking about covering the earth with data centers and supercomputers, 0:08:58.760 --> 0:09:02.480 leading to climate harm, healthcarms, the exacerbation of the fresh 0:09:02.520 --> 0:09:05.040 water crisis because these data centers need to be cooled 0:09:05.080 --> 0:09:09.800 with fresh water, and also in order to meet the 0:09:09.920 --> 0:09:13.920 data imperative for the size of these models. That is, 0:09:13.960 --> 0:09:17.640 when they start training on polluted data sets. That was 0:09:17.679 --> 0:09:20.560 not a norm at all in AI research. In fact, 0:09:20.720 --> 0:09:23.720 before Opening I started training on polluted data sets, the 0:09:24.000 --> 0:09:29.280 norm was actually shifting towards really small, extremely curated, and 0:09:29.360 --> 0:09:32.720 clean data sets. A lot of research was coming out 0:09:32.720 --> 0:09:37.600 at the time pre the CHATGBT moment where people realize 0:09:37.600 --> 0:09:40.520 you could actually get away with very tiny data sets 0:09:41.080 --> 0:09:46.160 if you prepared it correctly. But then Opening I shifted 0:09:46.280 --> 0:09:50.440 to let's use huge data sets, poor quality data sets, 0:09:50.720 --> 0:09:53.120 and that's why you end up having the need for 0:09:53.200 --> 0:09:56.560 content moderation, because when you're pumping a bunch of gunk 0:09:57.040 --> 0:09:59.120 into the models, then a bunch of gun comes out, 0:09:59.520 --> 0:10:02.920 and then you have to create content moderation filters to 0:10:03.040 --> 0:10:06.320 strip that gunk out before it reaches the user. And 0:10:06.320 --> 0:10:09.480 that involves humans, and that involves humans who then are 0:10:09.600 --> 0:10:12.920 left with a significant amount of trauma. 0:10:13.040 --> 0:10:15.560 There is a reason that we're talking about Sam Altman 0:10:15.679 --> 0:10:18.800 and open ai so much. Straight up. It's the same 0:10:18.920 --> 0:10:22.080 reason that people say chat GBT as a generic term 0:10:22.120 --> 0:10:25.240 for AI in general. Open Ai was the first company 0:10:25.280 --> 0:10:28.400 to break through and everyone copy them, and so the 0:10:28.440 --> 0:10:32.400 company culture at open Ai has influenced how other companies act, 0:10:32.640 --> 0:10:36.439 even Google and Microsoft, and then in turn, that's influenced 0:10:36.440 --> 0:10:39.559 how the public thinks about AI. And really a lot 0:10:39.600 --> 0:10:44.040 of our conversation about AI isn't about computer science. It's 0:10:44.040 --> 0:10:49.040 about culture, or maybe subculture, and that subculture is very 0:10:49.080 --> 0:10:52.440 interested in a fight between the good and evil use 0:10:52.559 --> 0:10:56.600 of AI. And it has this question that keeps coming 0:10:56.679 --> 0:11:00.400 up in the book what's your P? Doom? So I 0:11:01.000 --> 0:11:03.520 And before somebody listens to this and thinks, dexter, what 0:11:03.520 --> 0:11:07.520 did you just ask this person? P parentheses in the 0:11:07.520 --> 0:11:08.480 word doom. 0:11:08.640 --> 0:11:10.720 So I'm not a great fan of this phrase. It 0:11:10.760 --> 0:11:15.920 refers to probability of doom, and that is a shorthand 0:11:15.960 --> 0:11:22.880 within a particular community, ideological community that believes that AI 0:11:23.520 --> 0:11:26.880 has the potential to destroy humanity, and so probability of 0:11:26.960 --> 0:11:30.640 doom means the probability that humanity will be destroyed by 0:11:30.760 --> 0:11:33.160 artificial intelligence. The reason why I'm not a fan of 0:11:33.200 --> 0:11:36.880 it is because this ideology is predicated on the idea 0:11:37.520 --> 0:11:42.080 that we can fundamentally recreate human intelligence in computers, which 0:11:42.120 --> 0:11:45.880 is something that is still heavily scientifically debated. And this 0:11:45.920 --> 0:11:49.560 community also believes that once that happens, AI systems will 0:11:49.559 --> 0:11:55.559 develop their own consciousness or self motivation or self preservation, 0:11:55.720 --> 0:11:58.920 whatever it is that then makes them quote unquote go 0:11:59.200 --> 0:12:02.200 rogue and be unable to be controlled by the humans 0:12:02.280 --> 0:12:05.080 that originally program them. And that's what will lead to 0:12:05.280 --> 0:12:09.280 potential disaster of them just killing everyone or consuming all 0:12:09.320 --> 0:12:12.200 the resources such that most people have horrible lives, or 0:12:12.320 --> 0:12:14.959 keeping us as pets. You know, these are all scenarios, 0:12:14.960 --> 0:12:16.320 but the community talks about. 0:12:16.840 --> 0:12:19.120 When you say this is what the community talks about. 0:12:19.280 --> 0:12:22.280 There are people who like ask other people, like at 0:12:22.280 --> 0:12:24.800 a party or something, Yeah, so what's your P doom 0:12:24.840 --> 0:12:27.240 in somebody else? Yes, Oh mine's thirty five, And oh yeah, 0:12:27.360 --> 0:12:29.200 you know what I'm at about a seventy five. Recently, 0:12:29.200 --> 0:12:30.560 my P doom is seventy five. 0:12:30.640 --> 0:12:35.280 This happens, Yes, this happens. In addition to me fundamentally 0:12:35.320 --> 0:12:39.200 disagreeing with the ideological foundations of probability of doom, I 0:12:39.400 --> 0:12:44.600 also disagree with the veneer of rigor that a phrase 0:12:44.679 --> 0:12:48.680 like p doom ascribes to something that is inherently unrigorous. 0:12:49.360 --> 0:12:53.760 So they're putting mathematical values to something that is inherently illogical. 0:12:54.240 --> 0:12:56.439 There is no scientific evidence that we can point to 0:12:56.600 --> 0:13:00.199 that AI will go rogue, that it will do any 0:13:00.240 --> 0:13:03.160 of these things. It really is based on a belief, 0:13:03.320 --> 0:13:06.840 and over the last few years, especially within Silicon Valley, 0:13:06.880 --> 0:13:10.240 there's been the cultivation of what I call quasi religious 0:13:10.280 --> 0:13:14.240 movements around what AI is, what it will be, and 0:13:14.480 --> 0:13:18.280 ultimately how it will impact people. And we just talked 0:13:18.280 --> 0:13:21.079 about what most people call the Doomer ideology, and then 0:13:21.080 --> 0:13:24.359 there's the boomer ideology, which also believes that it's fundamentally 0:13:24.360 --> 0:13:28.080 possible to recreate human intelligence in computers, but for them 0:13:28.320 --> 0:13:32.240 this will be a positive civilizational transformation rather than a 0:13:32.240 --> 0:13:35.679 negative one. And both of them do not actually have 0:13:35.760 --> 0:13:37.560 scientific evidence one where or the other. They are talking 0:13:37.600 --> 0:13:41.520 about theoretical scenarios in projecting into the future based on 0:13:41.880 --> 0:13:45.000 their own conceptions of what human intelligence is and the 0:13:45.040 --> 0:13:50.000 idea that human intelligence is fundamentally computable, and they do 0:13:50.080 --> 0:13:54.520 not actually observe the real world harms or benefits of 0:13:54.559 --> 0:13:59.000 this technology as part of these philosophies, these ideologies. 0:13:59.000 --> 0:14:03.720 Okay, clarify here when people talk about boomers and doomers 0:14:03.760 --> 0:14:06.800 in the world of AI. Again, a boomer is someone 0:14:06.800 --> 0:14:10.840 who is really optimistic about AI. For example, believing that 0:14:10.920 --> 0:14:14.760 once we get artificial general intelligence, or basically an AI 0:14:14.880 --> 0:14:18.160 that's overall smarter than humans, that that will lead us 0:14:18.240 --> 0:14:22.800 to a utopia, prosperity, solving disease, fixing the climate, having 0:14:22.880 --> 0:14:25.640 a colony on the moon, food for everybody, all that 0:14:25.680 --> 0:14:29.600 sort of thing. A doomer is pessimistic about it. They 0:14:29.640 --> 0:14:32.440 think that AI will get smart and then suddenly go 0:14:32.560 --> 0:14:35.760 full sky net and kill everybody. So usually when we 0:14:35.800 --> 0:14:38.920 hear that there are two sides to an argument, we figure, Okay, 0:14:39.320 --> 0:14:42.360 the truth is probably somewhere in the middle there. But 0:14:42.800 --> 0:14:46.000 what if both sides are wrong? What if we're arguing 0:14:46.040 --> 0:14:50.040 about a question that doesn't even make sense? And this 0:14:50.120 --> 0:14:52.760 is one of the really interesting things I think about 0:14:52.960 --> 0:14:55.360 the process of reading the book. Right, it starts off 0:14:55.400 --> 0:14:58.400 with us as an inner office drama. There's some backstabbing here. 0:14:58.520 --> 0:15:01.040 There's a group of employees that really believes in this 0:15:01.160 --> 0:15:03.400 leader that goes over there and they're arguing and who's 0:15:03.440 --> 0:15:04.800 going to be the next leader, and do we need 0:15:04.840 --> 0:15:06.080 to get rid of this guy? Do we not? You 0:15:06.120 --> 0:15:08.240 need to get rid of this guy? But you've talked 0:15:08.280 --> 0:15:11.440 to so many people as you start to see how 0:15:11.480 --> 0:15:14.440 people are interacting with each other. You realize, Wait, this 0:15:14.560 --> 0:15:18.240 is an inner office drama or personalities coming up against 0:15:18.240 --> 0:15:22.640 each other. This is people who fundamentally believe something fairly deeply, 0:15:23.360 --> 0:15:25.960 and those things are really clashing against each other. 0:15:26.200 --> 0:15:29.680 Yeah. The shocking thing was as I was interviewing people 0:15:29.720 --> 0:15:32.200 who I identified as part of the Boomer group or 0:15:32.240 --> 0:15:34.800 part of the Doomer group. I mean the boomers, like 0:15:34.840 --> 0:15:38.680 their eyes would light up when they were talking about 0:15:38.800 --> 0:15:44.720 this potential future where prosperity was abundant and everything would 0:15:44.720 --> 0:15:48.600 be perfect, and the doomers, like I spoke to people 0:15:48.600 --> 0:15:52.560 whose voices were quivering with anxiety. This was a genuine emotional, 0:15:53.160 --> 0:15:57.240 visceral reaction to the ideas. Wow, we only had a 0:15:57.240 --> 0:16:01.240 few years left on this earth potentially if we did 0:16:01.280 --> 0:16:04.880 not figure out how to get a handle on this 0:16:04.960 --> 0:16:08.920 technology and make it go well instead of badly. And 0:16:09.280 --> 0:16:13.080 that's when I began to understand more deeply. Oh, there 0:16:13.120 --> 0:16:18.080 are so many more layers to all of the headlines 0:16:18.120 --> 0:16:21.120 that we see about this company, about this technology, and 0:16:21.600 --> 0:16:24.400 about the dramatic firing and rehiring of Sam Woman. There's 0:16:24.440 --> 0:16:29.400 so many deeper spiritual layers behind the clashing that's happening 0:16:29.440 --> 0:16:31.280 to shape this technology. 0:16:31.080 --> 0:16:34.120 But really, what is this spiritual fight that we're trying 0:16:34.160 --> 0:16:48.240 to have here? More on that after the break. So 0:16:48.520 --> 0:16:52.080 it's interesting that people sometimes compare Sam Altman, the head 0:16:52.120 --> 0:16:54.840 of open Ai, to Steve Jobs. You say in the 0:16:54.880 --> 0:16:58.320 book that in some ways he's this generational talent, but 0:16:58.640 --> 0:17:00.760 you also explain that there's a lot of people who 0:17:00.800 --> 0:17:04.560 really just don't trust him, that he's polarizing, because obviously 0:17:04.600 --> 0:17:07.080 there's a way in which I mean Steve Jobs is 0:17:07.119 --> 0:17:10.239 also a polarizing figure too. There's people who think he's 0:17:10.280 --> 0:17:12.560 an absolute genius. There's people who think he's a jerk 0:17:12.560 --> 0:17:15.560 who just stole ideas, and there's people who think he's 0:17:15.600 --> 0:17:18.560 a genius and he's a jerk. Yeah, yeah, right, But 0:17:18.600 --> 0:17:21.440 then if you think here about what are the stakes 0:17:21.480 --> 0:17:23.560 of what Steve Jobs was doing? If you want to 0:17:23.560 --> 0:17:26.800 give him all the credit, yeah, is what was he doing? 0:17:26.800 --> 0:17:30.480 What's the mission? Make beautiful products, make very easy to 0:17:30.560 --> 0:17:34.680 use computers great? What are the stakes for open ai? Actually, 0:17:34.760 --> 0:17:36.479 hold on, let me not even try to answer this. 0:17:36.800 --> 0:17:39.639 What is open AI's mission because that's something that seems 0:17:39.640 --> 0:17:43.600 to change through the book. So what is open ai 0:17:43.680 --> 0:17:44.400 trying to do here? 0:17:44.920 --> 0:17:49.480 Yeah? So their mission, which on paper has never changed, 0:17:49.760 --> 0:17:53.720 is to ensure artificial general intelligence benefits all of humanity. 0:17:53.840 --> 0:17:57.520 That's the direct quote. The challenge is that each of 0:17:57.560 --> 0:18:02.359 these components of its mission are extremely ill defined. I mean, 0:18:02.440 --> 0:18:04.680 there used to be a joke with an open eye. 0:18:04.720 --> 0:18:08.280 If you ask thirteen researchers at the company what artificial 0:18:08.320 --> 0:18:13.080 general intelligence is, you'll get fifteen answers. And it's pretty 0:18:13.119 --> 0:18:16.400 true for all the components of the mission. So what's 0:18:16.440 --> 0:18:20.040 happened over the course of the organization's history it originally 0:18:20.040 --> 0:18:22.359 started as a nonprofit, now it's one of the most 0:18:22.359 --> 0:18:27.479 capitalistic companies in Silicon Valley history, is that different people 0:18:27.760 --> 0:18:31.639 interpret the mission in a fundamentally different way. The Boomers 0:18:31.760 --> 0:18:34.639 interpret benefiting all of humanity as build this technology as 0:18:34.640 --> 0:18:37.239 fast as possible and unleash it onto the world as 0:18:37.280 --> 0:18:40.960 quickly as possible. The Doomers interpret as build this technology 0:18:40.960 --> 0:18:44.520 as fast as possible and hold onto the technology so 0:18:44.600 --> 0:18:47.360 that we have a lead time to do more research 0:18:47.400 --> 0:18:49.840 on it before bad actors have a chance to do 0:18:49.880 --> 0:18:52.480 research on it as well. And then there are plenty 0:18:52.520 --> 0:18:54.640 of other people who are not necessarily in the boom 0:18:54.680 --> 0:18:57.920 or a dumer category that are just regular tech company, 0:18:58.160 --> 0:19:00.879 people who came from Facebook, who it came from Google, 0:19:00.920 --> 0:19:05.119 who came from Microsoft, who are just like the mission 0:19:05.160 --> 0:19:08.240 for benefiting all of humanity is building products that people 0:19:08.240 --> 0:19:11.439 want to pay for. There's such a vast range of 0:19:11.960 --> 0:19:16.320 interpretations that essentially all the mission does is it just 0:19:16.359 --> 0:19:18.520 allows people to put a mirror up to themselves and 0:19:18.560 --> 0:19:21.520 say what is it that I want and to make 0:19:21.560 --> 0:19:23.919 themselves the protagonist of their own story and say what 0:19:23.960 --> 0:19:26.320 I want is the most beneficial for humanity. So that's 0:19:26.880 --> 0:19:30.200 my interpretation of what Opening I should be doing, and 0:19:30.240 --> 0:19:33.160 that is part of the reason why Opening I has 0:19:33.359 --> 0:19:37.000 had so much drama through its history, because no one 0:19:37.000 --> 0:19:42.199 can ever agree on the most fundamental building block of 0:19:42.240 --> 0:19:46.320 the company. No one can agree on what direction they 0:19:46.359 --> 0:19:47.640 should actually be going. 0:19:48.000 --> 0:19:51.840 Speaking of benefiting all of humanity, a couple episodes ago, 0:19:52.280 --> 0:19:55.159 we did an episode about the impact of AI on 0:19:55.200 --> 0:19:58.680 the environment, climate change, things like that. But then there's 0:19:58.760 --> 0:20:02.720 this kind of background for anybody who's really interested in AI, 0:20:02.800 --> 0:20:06.240 really kind of AI promoter. There's this counterclaim that, Okay, 0:20:06.280 --> 0:20:09.680 any bad stuff we do to the environment, Ai'll fix it. Yeah, 0:20:09.800 --> 0:20:13.760 AI can fix climate change. Yeah, you've run up against 0:20:13.800 --> 0:20:17.320 this claim in person more than I have. Have you 0:20:17.359 --> 0:20:19.440 been able to make any sense of that? What's the 0:20:19.520 --> 0:20:21.760 argument here that AI is going to fix climate change 0:20:21.840 --> 0:20:22.880 or AI can help there? 0:20:23.240 --> 0:20:29.840 The most charitable relaying of the argument is for people 0:20:29.920 --> 0:20:34.520 who believe that human intelligence is fundamentally computable, and therefore 0:20:34.600 --> 0:20:36.879 if you have enough data and you have enough compute, 0:20:36.920 --> 0:20:40.520 you will inevitably be able to recreate it and create 0:20:40.760 --> 0:20:44.560 so called artificial general intelligence. Then you should be able 0:20:44.600 --> 0:20:49.359 to solve any problem at that point, because the challenge 0:20:49.840 --> 0:20:53.919 with our as humans, our inability to deal with the 0:20:53.920 --> 0:20:59.840 climate crisis is a lack of cooperation and digital intelligence. 0:20:59.840 --> 0:21:02.320 They won't have egos, they won't be like clashing against 0:21:02.320 --> 0:21:04.840 each other, so the saying goes, and so they won't 0:21:04.840 --> 0:21:08.880 have any issue cooperating, and they also won't have any 0:21:09.000 --> 0:21:13.399 lack of ideas and ability to experiment, develop new energy 0:21:13.400 --> 0:21:17.320 storage solutions, develop new forms of renewable energy, and so 0:21:17.400 --> 0:21:19.679 on and so forth. So that is the kind of 0:21:19.720 --> 0:21:25.000 most charitable version of why artificial general intelligence could fundamentally 0:21:25.040 --> 0:21:29.879 solve climate change. My critique is, again, we don't have 0:21:29.960 --> 0:21:33.560 scientific evidence that AGI will ever come to pass, and 0:21:33.640 --> 0:21:39.639 so we are basically trying to justify current day vast 0:21:39.800 --> 0:21:43.240 environmental harms and the acceleration of the climate crisis with 0:21:43.320 --> 0:21:47.400 a speculative possibility that it might one day be able 0:21:47.440 --> 0:21:50.960 to go away. And so we're essentially trying to cover 0:21:51.200 --> 0:21:57.359 up real scientific evidence of present day reality with a 0:21:57.600 --> 0:22:01.240 spiritual belief that it'll all be okay in the end. 0:22:01.640 --> 0:22:04.000 Listen. I would go further. And I mean my issue 0:22:04.000 --> 0:22:07.880 with this is even if an AI agent, your chatbot, 0:22:07.880 --> 0:22:10.520 can spit out the answer, Hey, here's what to do, 0:22:10.520 --> 0:22:13.480 do this, do this, and do this, and climate change 0:22:13.520 --> 0:22:16.600 will grind to a halt. What if we're not interested 0:22:16.640 --> 0:22:18.560 in listening, you know what I mean? 0:22:18.920 --> 0:22:19.520 Yeah? Right? 0:22:19.800 --> 0:22:21.520 You could get on chat YOUBT, you can get on 0:22:21.560 --> 0:22:24.720 claud you can get on Gemini or whatever and say, hey, hey, 0:22:24.760 --> 0:22:28.320 AI friend, I'm I'm feeling really sick. I'm eating all 0:22:28.359 --> 0:22:30.240 of this cake and I really don't feel good. What 0:22:30.280 --> 0:22:32.840 should I do? And I'll say, hey, buddy, stop eating cake. 0:22:33.240 --> 0:22:34.359 I'm gonna keep eating cake. 0:22:34.440 --> 0:22:34.520 Yo. 0:22:34.960 --> 0:22:37.720 You don't actually have to listen. The computer can't make 0:22:37.760 --> 0:22:40.800 you listen, even if it has the answer. And here 0:22:40.840 --> 0:22:44.000 in the States particularly, we've got the data. There are 0:22:44.040 --> 0:22:46.800 some ideas on things we could do to reduce the 0:22:46.880 --> 0:22:48.280 impact on the environment. 0:22:48.400 --> 0:22:51.320 We're just not doing them exactly. I think this is 0:22:51.320 --> 0:22:53.600 one of the things that has always broken down in 0:22:53.640 --> 0:22:56.760 these theoretical future AGI arguments of whether it's going to 0:22:56.800 --> 0:23:01.240 be fundamentally positively or negatively transformative. There's never a clear 0:23:01.320 --> 0:23:05.280 articulation of how it's operationalized in the physical world. Are 0:23:05.359 --> 0:23:07.960 they going to have robot bodies, are they going to 0:23:07.960 --> 0:23:12.639 be mining the earth and developing and cultivating these new 0:23:12.760 --> 0:23:17.400 energy storage solutions, or are they directing humans to do that, 0:23:18.040 --> 0:23:20.760 at which point we still run into all of the 0:23:20.800 --> 0:23:24.680 same problems that we've always right pad, which is humans 0:23:24.760 --> 0:23:27.320 don't listen. So I think you're hitting upon one of 0:23:27.400 --> 0:23:32.119 the core weaknesses of just many of the arguments in 0:23:32.160 --> 0:23:37.440 this community is they do not actually acknowledge the social, political, 0:23:37.520 --> 0:23:44.760 and economic aspects of the way that technology ultimately impacts society. 0:23:45.160 --> 0:23:47.960 Like it's not just you have a technical capability and 0:23:48.000 --> 0:23:52.280 everything is suddenly solved. There's so many more layers to 0:23:52.800 --> 0:23:56.120 how the capability then translates into real world impact. 0:23:58.000 --> 0:24:01.919 Yeah, I'm sympathetic to some of the ideas that are 0:24:01.960 --> 0:24:05.399 put forth, like AI could cure diseases. Hey, look, you 0:24:05.400 --> 0:24:07.639