Latest / kill switch / boomers, doomers, and the new empire
Transcript
- 0:11Speaker 1: Empire of Ai. You're using the word empire here. Yeah,
- 0:14Speaker 1: Can you tell me a little bit more about that decision?
- 0:16Speaker 2: Yeah, So the term empire in the title is specifically
- 0:21Speaker 2: an argument that I make in the book that we
- 0:24Speaker 2: really need to start thinking of companies like open Ai
- 0:26Speaker 2: as new forms of empire.
- 0:29Speaker 1: Karen Howe is a journalist and the author of a
- 0:32Speaker 1: new book called Empire of Ai Dreams and Nightmares and
- 0:36Speaker 1: Sam Altman's Open Ai, which just came out last month.
- 0:39Speaker 1: In case you're not familiar with her work, she herself
- 0:42Speaker 1: is a data engineer, and she's also the first person
- 0:45Speaker 1: to ever really cover open Ai, the company that makes
- 0:48Speaker 1: chat gpt. On the surface, the book is a story
- 0:51Speaker 1: of the rise of open Ai and the story of
- 0:53Speaker 1: its founder, Sam Altman, and if you like drama, let
- 0:57Speaker 1: me tell you there's a lot of it. Someone could
- 0:59Speaker 1: definitely make a Netflix series about this. And they are
- 1:02Speaker 1: also comparisons to be made between Sam Altman and Steve Jobs,
- 1:07Speaker 1: and we will get to that. But before we do anything,
- 1:10Speaker 1: we should talk about a word that a lot of
- 1:12Speaker 1: the reviews I've seen are kind of glossing over, which
- 1:15Speaker 1: is weird because it's the first word in the title empire, and.
- 1:20Speaker 2: The reason is four different features that empires of AI
- 1:23Speaker 2: share with empires of old. The first one is that
- 1:26Speaker 2: they lay claim to resources that are not their own,
- 1:28Speaker 2: but they reinterpret the rules to suggest that those resources
- 1:32Speaker 2: were always their own. That refers to how these companies
- 1:35Speaker 2: just scrape the data on the Internet, same with the
- 1:37Speaker 2: intellectual property that these companies just take, like the work
- 1:41Speaker 2: of artists, the work of writers, the work of creators.
- 1:43Speaker 2: The second feature of empires is that they engage in
- 1:46Speaker 2: a lot of labor exploitation, and that refers not just
- 1:49Speaker 2: to the fact that these companies they contract a lot
- 1:52Speaker 2: of workers, often in the Global South or other economically
- 1:55Speaker 2: vulnerable communities in the Global North, and pay them extraordinarily
- 1:59Speaker 2: small amounts of money to do extremely exploitative work data annotation,
- 2:04Speaker 2: data cleaning, data preparation, content moderation, where workers are left
- 2:09Speaker 2: with the same level of trauma that social media content
- 2:12Speaker 2: moderators were left with. It also refers to the fact
- 2:15Speaker 2: that these companies are inherently building labor automating technologies. So
- 2:21Speaker 2: opening eyes definition of artificial general intelligence is highly autonomous
- 2:25Speaker 2: systems that outperform humans at most economically valuable work and
- 2:31Speaker 2: so they're explicitly saying we're going to automate away the
- 2:35Speaker 2: things that people usually get paid for, and that in
- 2:38Speaker 2: and of itself is a labor exploitation. So there's labor
- 2:40Speaker 2: exploitation happening on the way in and happening on the
- 2:43Speaker 2: way out.
- 2:44Speaker 1: Karen is not kidding. In the book, she travels all
- 2:47Speaker 1: over the world and speaks to people who work in
- 2:50Speaker 1: all levels of the AI industry. And what you've found
- 2:54Speaker 1: was that countries that were undergoing some kind of economic
- 2:57Speaker 1: crisis were being targeted by the industry as places to
- 3:00Speaker 1: hire workers who train AI models to understand what kind
- 3:03Speaker 1: of stuff to allow and what to block. This is
- 3:07Speaker 1: work that only a human can do. So who's this
- 3:11Speaker 1: empire for? And where do you and I fit in here?
- 3:21Speaker 1: I'm afraid Kaleidoscope and iHeart podcasts. This is kill switch.
- 3:32Speaker 1: I'm doctor Thomas.
- 3:34Speaker 3: I'm sorry, I'm goodbye.
- 4:03Speaker 1: Getting back to Karen's definitions of colonial empires, the first
- 4:07Speaker 1: is that they change the rules so that when they
- 4:09Speaker 1: find something like oil in the ground or art and
- 4:12Speaker 1: literature on the internet, they can say that it belongs
- 4:15Speaker 1: to them. The second is that they exploit people for
- 4:18Speaker 1: their labor. The third is that they control what people
- 4:21Speaker 1: can know about them. In the case of the AI industry,
- 4:24Speaker 1: this can look like companies just finding the top researchers
- 4:28Speaker 1: who are working at universities, paying them lots of money
- 4:31Speaker 1: to drop their university job and come to the company,
- 4:34Speaker 1: and then censoring anything that those researchers write that is
- 4:37Speaker 1: critical of their technology or of how they treat the environment. Again,
- 4:42Speaker 1: controlling what the public knows about them. But I think
- 4:46Speaker 1: it's the fourth element where things start to really get interesting.
- 4:49Speaker 2: And then the last future is that empires always have
- 4:52Speaker 2: this narrative that there are good empires and their evil empires.
- 4:57Speaker 2: So they the good empire need to be empire in
- 5:00Speaker 2: the first place and engage in all this resource extraction
- 5:02Speaker 2: and labor exploitation because that is what will make them
- 5:06Speaker 2: strong enough to beat back the evil empire. So back
- 5:08Speaker 2: in the day and during European colonialism, the British Empire
- 5:11Speaker 2: would say that they were better than the Dutch Empire.
- 5:13Speaker 2: The French Empire would say they were better than the
- 5:14Speaker 2: British Empire. And part of being the good empire is
- 5:19Speaker 2: that what they were ultimately doing was civilizing the world.
- 5:24Speaker 2: They were bringing progress and modernity to everyone, and they
- 5:28Speaker 2: were giving humanity the chance to enter heaven instead of hell.
- 5:33Speaker 2: And this is literally the language that AI people use
- 5:39Speaker 2: these days. They talk about heaven and hell. They talk
- 5:42Speaker 2: about recreating God and about bringing the next era of
- 5:48Speaker 2: civilization into being.
- 5:51Speaker 1: Yeah, and I mean Sam Altman twenty twenty three going
- 5:54Speaker 1: before Congress and saying, look, if we don't put a
- 5:58Speaker 1: whole bunch of resources in AA, guess who will China?
- 6:01Speaker 1: Would you like China to win? I don't think you do.
- 6:04Speaker 1: I know that you're scared of China. Let's play nice
- 6:07Speaker 1: here so that you know, play nice specifically with our industry.
- 6:10Speaker 1: Also play nice with my company so that we can
- 6:13Speaker 1: come out on top, which again is very similar to
- 6:17Speaker 1: how an empire justifies itself, which is, if we don't
- 6:22Speaker 1: do this, the bad guys will, such as you lay
- 6:25Speaker 1: on the book open AI really starts out by saying, listen,
- 6:29Speaker 1: AI is going to happen. AJI is going to happen.
- 6:33Speaker 1: It needs to be the good guys. We're the good guys.
- 6:36Speaker 1: And I think that's something that you do in the
- 6:37Speaker 1: book is I think we've gotten really used to this
- 6:41Speaker 1: idea that everything was inevitable this all was going to happen,
- 6:44Speaker 1: and the stuff that happens tomorrow it was always going
- 6:47Speaker 1: to happen, And the stuff that happens three years from now,
- 6:49Speaker 1: it was always going to happen. But you put it
- 6:52Speaker 1: together that none of this is actually inevitable, and not
- 6:56Speaker 1: only that, it's decisions are being made, and actually a
- 6:59Speaker 1: lot of these decisions are being made by a pretty
- 7:03Speaker 1: small handful of people in Silicon Valley.
- 7:06Speaker 2: Absolutely, this is definitely one of the core messages that
- 7:09Speaker 2: I hope people can take away from the book is
- 7:12Speaker 2: technology is very much a product of human choices, and
- 7:15Speaker 2: it just so happens that AI has been the product
- 7:18Speaker 2: of a very small handful of people's choices who have
- 7:21Speaker 2: a very what I would argue, narrow world view, a
- 7:25Speaker 2: narrow view of how the world works now and how
- 7:28Speaker 2: it should continue to work. And if I had to
- 7:31Speaker 2: point to one decision that was not inevitable, but that
- 7:35Speaker 2: Opening Eye sort of ushered in with the shape of
- 7:37Speaker 2: AI today, it is the fact that they decided to
- 7:41Speaker 2: scale this technology aggressively, and that happened because open Ai
- 7:47Speaker 2: early on identified that they had to be number one
- 7:51Speaker 2: based on this. There's the evil Empire out there, and
- 7:53Speaker 2: we the Good Empire, have to race aggressively. And they
- 7:58Speaker 2: realized that the fact fastest and easiest way to get
- 8:03Speaker 2: to number one was by taking existing techniques within the
- 8:06Speaker 2: AI research field and blowing it up with an unprecedented
- 8:10Speaker 2: amount of data and an unprecedented amount of computational resources.
- 8:14Speaker 2: So they realized, if we can build the largest supercomputers
- 8:16Speaker 2: in the world, we will have the best chance at
- 8:21Speaker 2: dominating this technology. All of a sudden, all of the
- 8:25Speaker 2: tech companies saw what opening I did and when we
- 8:28Speaker 2: want to do that too, And suddenly you see a
- 8:31Speaker 2: step change in the sheer amount of resources that are
- 8:34Speaker 2: now going into developing this technology, and it very much
- 8:37Speaker 2: becomes a scale at all costs paradigm for AI development.
- 8:42Speaker 2: This very specific scaling decision is what leads to significant
- 8:48Speaker 2: amounts of labor exploitation and is what leads to significant
- 8:51Speaker 2: amounts of environmental harm, because that is when you start
- 8:55Speaker 2: talking about covering the earth with data centers and supercomputers,
- 8:58Speaker 2: leading to climate harm, healthcarms, the exacerbation of the fresh
- 9:02Speaker 2: water crisis because these data centers need to be cooled
- 9:05Speaker 2: with fresh water, and also in order to meet the
- 9:09Speaker 2: data imperative for the size of these models. That is,
- 9:13Speaker 2: when they start training on polluted data sets. That was
- 9:17Speaker 2: not a norm at all in AI research. In fact,
- 9:20Speaker 2: before Opening I started training on polluted data sets, the
- 9:24Speaker 2: norm was actually shifting towards really small, extremely curated, and
- 9:29Speaker 2: clean data sets. A lot of research was coming out
- 9:32Speaker 2: at the time pre the CHATGBT moment where people realize
- 9:37Speaker 2: you could actually get away with very tiny data sets
- 9:41Speaker 2: if you prepared it correctly. But then Opening I shifted
- 9:46Speaker 2: to let's use huge data sets, poor quality data sets,
- 9:50Speaker 2: and that's why you end up having the need for
- 9:53Speaker 2: content moderation, because when you're pumping a bunch of gunk
- 9:57Speaker 2: into the models, then a bunch of gun comes out,
- 9:59Speaker 2: and then you have to create content moderation filters to
- 10:03Speaker 2: strip that gunk out before it reaches the user. And
- 10:06Speaker 2: that involves humans, and that involves humans who then are
- 10:09Speaker 2: left with a significant amount of trauma.
- 10:13Speaker 1: There is a reason that we're talking about Sam Altman
- 10:15Speaker 1: and open ai so much. Straight up. It's the same
- 10:18Speaker 1: reason that people say chat GBT as a generic term
- 10:22Speaker 1: for AI in general. Open Ai was the first company
- 10:25Speaker 1: to break through and everyone copy them, and so the
- 10:28Speaker 1: company culture at open Ai has influenced how other companies act,
- 10:32Speaker 1: even Google and Microsoft, and then in turn, that's influenced
- 10:36Speaker 1: how the public thinks about AI. And really a lot
- 10:39Speaker 1: of our conversation about AI isn't about computer science. It's
- 10:44Speaker 1: about culture, or maybe subculture, and that subculture is very
- 10:49Speaker 1: interested in a fight between the good and evil use
- 10:52Speaker 1: of AI. And it has this question that keeps coming
- 10:56Speaker 1: up in the book what's your P? Doom? So I
- 11:01Speaker 1: And before somebody listens to this and thinks, dexter, what
- 11:03Speaker 1: did you just ask this person? P parentheses in the
- 11:07Speaker 1: word doom.
- 11:08Speaker 2: So I'm not a great fan of this phrase. It
- 11:10Speaker 2: refers to probability of doom, and that is a shorthand
- 11:15Speaker 2: within a particular community, ideological community that believes that AI
- 11:23Speaker 2: has the potential to destroy humanity, and so probability of
- 11:26Speaker 2: doom means the probability that humanity will be destroyed by
- 11:30Speaker 2: artificial intelligence. The reason why I'm not a fan of
- 11:33Speaker 2: it is because this ideology is predicated on the idea
- 11:37Speaker 2: that we can fundamentally recreate human intelligence in computers, which
- 11:42Speaker 2: is something that is still heavily scientifically debated. And this
- 11:45Speaker 2: community also believes that once that happens, AI systems will
- 11:49Speaker 2: develop their own consciousness or self motivation or self preservation,
- 11:55Speaker 2: whatever it is that then makes them quote unquote go
- 11:59Speaker 2: rogue and be unable to be controlled by the humans
- 12:02Speaker 2: that originally program them. And that's what will lead to
- 12:05Speaker 2: potential disaster of them just killing everyone or consuming all
- 12:09Speaker 2: the resources such that most people have horrible lives, or
- 12:12Speaker 2: keeping us as pets. You know, these are all scenarios,
- 12:14Speaker 2: but the community talks about.
- 12:16Speaker 1: When you say this is what the community talks about.
- 12:19Speaker 1: There are people who like ask other people, like at
- 12:22Speaker 1: a party or something, Yeah, so what's your P doom
- 12:24Speaker 1: in somebody else? Yes, Oh mine's thirty five, And oh yeah,
- 12:27Speaker 1: you know what I'm at about a seventy five. Recently,
- 12:29Speaker 1: my P doom is seventy five.
- 12:30Speaker 2: This happens, Yes, this happens. In addition to me fundamentally
- 12:35Speaker 2: disagreeing with the ideological foundations of probability of doom, I
- 12:39Speaker 2: also disagree with the veneer of rigor that a phrase
- 12:44Speaker 2: like p doom ascribes to something that is inherently unrigorous.
- 12:49Speaker 2: So they're putting mathematical values to something that is inherently illogical.
- 12:54Speaker 2: There is no scientific evidence that we can point to
- 12:56Speaker 2: that AI will go rogue, that it will do any
- 13:00Speaker 2: of these things. It really is based on a belief,
- 13:03Speaker 2: and over the last few years, especially within Silicon Valley,
- 13:06Speaker 2: there's been the cultivation of what I call quasi religious
- 13:10Speaker 2: movements around what AI is, what it will be, and
- 13:14Speaker 2: ultimately how it will impact people. And we just talked
- 13:18Speaker 2: about what most people call the Doomer ideology, and then
- 13:21Speaker 2: there's the boomer ideology, which also believes that it's fundamentally
- 13:24Speaker 2: possible to recreate human intelligence in computers, but for them
- 13:28Speaker 2: this will be a positive civilizational transformation rather than a
- 13:32Speaker 2: negative one. And both of them do not actually have
- 13:35Speaker 2: scientific evidence one where or the other. They are talking
- 13:37Speaker 2: about theoretical scenarios in projecting into the future based on
- 13:41Speaker 2: their own conceptions of what human intelligence is and the
- 13:45Speaker 2: idea that human intelligence is fundamentally computable, and they do
- 13:50Speaker 2: not actually observe the real world harms or benefits of
- 13:54Speaker 2: this technology as part of these philosophies, these ideologies.
- 13:59Speaker 1: Okay, clarify here when people talk about boomers and doomers
- 14:03Speaker 1: in the world of AI. Again, a boomer is someone
- 14:06Speaker 1: who is really optimistic about AI. For example, believing that
- 14:10Speaker 1: once we get artificial general intelligence, or basically an AI
- 14:14Speaker 1: that's overall smarter than humans, that that will lead us
- 14:18Speaker 1: to a utopia, prosperity, solving disease, fixing the climate, having
- 14:22Speaker 1: a colony on the moon, food for everybody, all that
- 14:25Speaker 1: sort of thing. A doomer is pessimistic about it. They
- 14:29Speaker 1: think that AI will get smart and then suddenly go
- 14:32Speaker 1: full sky net and kill everybody. So usually when we
- 14:35Speaker 1: hear that there are two sides to an argument, we figure, Okay,
- 14:39Speaker 1: the truth is probably somewhere in the middle there. But
- 14:42Speaker 1: what if both sides are wrong? What if we're arguing
- 14:46Speaker 1: about a question that doesn't even make sense? And this
- 14:50Speaker 1: is one of the really interesting things I think about
- 14:52Speaker 1: the process of reading the book. Right, it starts off
- 14:55Speaker 1: with us as an inner office drama. There's some backstabbing here.
- 14:58Speaker 1: There's a group of employees that really believes in this
- 15:01Speaker 1: leader that goes over there and they're arguing and who's
- 15:03Speaker 1: going to be the next leader, and do we need
- 15:04Speaker 1: to get rid of this guy? Do we not? You
- 15:06Speaker 1: need to get rid of this guy? But you've talked
- 15:08Speaker 1: to so many people as you start to see how
- 15:11Speaker 1: people are interacting with each other. You realize, Wait, this
- 15:14Speaker 1: is an inner office drama or personalities coming up against
- 15:18Speaker 1: each other. This is people who fundamentally believe something fairly deeply,
- 15:23Speaker 1: and those things are really clashing against each other.
- 15:26Speaker 2: Yeah. The shocking thing was as I was interviewing people
- 15:29Speaker 2: who I identified as part of the Boomer group or
- 15:32Speaker 2: part of the Doomer group. I mean the boomers, like
- 15:34Speaker 2: their eyes would light up when they were talking about
- 15:38Speaker 2: this potential future where prosperity was abundant and everything would
- 15:44Speaker 2: be perfect, and the doomers, like I spoke to people
- 15:48Speaker 2: whose voices were quivering with anxiety. This was a genuine emotional,
- 15:53Speaker 2: visceral reaction to the ideas. Wow, we only had a
- 15:57Speaker 2: few years left on this earth potentially if we did
- 16:01Speaker 2: not figure out how to get a handle on this
- 16:04Speaker 2: technology and make it go well instead of badly. And
- 16:09Speaker 2: that's when I began to understand more deeply. Oh, there
- 16:13Speaker 2: are so many more layers to all of the headlines
- 16:18Speaker 2: that we see about this company, about this technology, and
- 16:21Speaker 2: about the dramatic firing and rehiring of Sam Woman. There's
- 16:24Speaker 2: so many deeper spiritual layers behind the clashing that's happening
- 16:29Speaker 2: to shape this technology.
- 16:31Speaker 1: But really, what is this spiritual fight that we're trying
- 16:34Speaker 1: to have here? More on that after the break. So
- 16:48Speaker 1: it's interesting that people sometimes compare Sam Altman, the head
- 16:52Speaker 1: of open Ai, to Steve Jobs. You say in the
- 16:54Speaker 1: book that in some ways he's this generational talent, but
- 16:58Speaker 1: you also explain that there's a lot of people who
- 17:00Speaker 1: really just don't trust him, that he's polarizing, because obviously
- 17:04Speaker 1: there's a way in which I mean Steve Jobs is
- 17:07Speaker 1: also a polarizing figure too. There's people who think he's
- 17:10Speaker 1: an absolute genius. There's people who think he's a jerk
- 17:12Speaker 1: who just stole ideas, and there's people who think he's
- 17:15Speaker 1: a genius and he's a jerk. Yeah, yeah, right, But
- 17:18Speaker 1: then if you think here about what are the stakes
- 17:21Speaker 1: of what Steve Jobs was doing? If you want to
- 17:23Speaker 1: give him all the credit, yeah, is what was he doing?
- 17:26Speaker 1: What's the mission? Make beautiful products, make very easy to
- 17:30Speaker 1: use computers great? What are the stakes for open ai? Actually,
- 17:34Speaker 1: hold on, let me not even try to answer this.
- 17:36Speaker 1: What is open AI's mission because that's something that seems
- 17:39Speaker 1: to change through the book. So what is open ai
- 17:43Speaker 1: trying to do here?
- 17:44Speaker 2: Yeah? So their mission, which on paper has never changed,
- 17:49Speaker 2: is to ensure artificial general intelligence benefits all of humanity.
- 17:53Speaker 2: That's the direct quote. The challenge is that each of
- 17:57Speaker 2: these components of its mission are extremely ill defined. I mean,
- 18:02Speaker 2: there used to be a joke with an open eye.
- 18:04Speaker 2: If you ask thirteen researchers at the company what artificial
- 18:08Speaker 2: general intelligence is, you'll get fifteen answers. And it's pretty
- 18:13Speaker 2: true for all the components of the mission. So what's
- 18:16Speaker 2: happened over the course of the organization's history it originally
- 18:20Speaker 2: started as a nonprofit, now it's one of the most
- 18:22Speaker 2: capitalistic companies in Silicon Valley history, is that different people
- 18:27Speaker 2: interpret the mission in a fundamentally different way. The Boomers
- 18:31Speaker 2: interpret benefiting all of humanity as build this technology as
- 18:34Speaker 2: fast as possible and unleash it onto the world as
- 18:37Speaker 2: quickly as possible. The Doomers interpret as build this technology
- 18:40Speaker 2: as fast as possible and hold onto the technology so
- 18:44Speaker 2: that we have a lead time to do more research
- 18:47Speaker 2: on it before bad actors have a chance to do
- 18:49Speaker 2: research on it as well. And then there are plenty
- 18:52Speaker 2: of other people who are not necessarily in the boom
- 18:54Speaker 2: or a dumer category that are just regular tech company,
- 18:58Speaker 2: people who came from Facebook, who it came from Google,
- 19:00Speaker 2: who came from Microsoft, who are just like the mission
- 19:05Speaker 2: for benefiting all of humanity is building products that people
- 19:08Speaker 2: want to pay for. There's such a vast range of
- 19:11Speaker 2: interpretations that essentially all the mission does is it just
- 19:16Speaker 2: allows people to put a mirror up to themselves and
- 19:18Speaker 2: say what is it that I want and to make
- 19:21Speaker 2: themselves the protagonist of their own story and say what
- 19:23Speaker 2: I want is the most beneficial for humanity. So that's
- 19:26Speaker 2: my interpretation of what Opening I should be doing, and
- 19:30Speaker 2: that is part of the reason why Opening I has
- 19:33Speaker 2: had so much drama through its history, because no one
- 19:37Speaker 2: can ever agree on the most fundamental building block of
- 19:42Speaker 2: the company. No one can agree on what direction they
- 19:46Speaker 2: should actually be going.
- 19:48Speaker 1: Speaking of benefiting all of humanity, a couple episodes ago,
- 19:52Speaker 1: we did an episode about the impact of AI on
- 19:55Speaker 1: the environment, climate change, things like that. But then there's
- 19:58Speaker 1: this kind of background for anybody who's really interested in AI,
- 20:02Speaker 1: really kind of AI promoter. There's this counterclaim that, Okay,
- 20:06Speaker 1: any bad stuff we do to the environment, Ai'll fix it. Yeah,
- 20:09Speaker 1: AI can fix climate change. Yeah, you've run up against
- 20:13Speaker 1: this claim in person more than I have. Have you
- 20:17Speaker 1: been able to make any sense of that? What's the
- 20:19Speaker 1: argument here that AI is going to fix climate change
- 20:21Speaker 1: or AI can help there?
- 20:23Speaker 2: The most charitable relaying of the argument is for people
- 20:29Speaker 2: who believe that human intelligence is fundamentally computable, and therefore
- 20:34Speaker 2: if you have enough data and you have enough compute,
- 20:36Speaker 2: you will inevitably be able to recreate it and create
- 20:40Speaker 2: so called artificial general intelligence. Then you should be able
- 20:44Speaker 2: to solve any problem at that point, because the challenge
- 20:49Speaker 2: with our as humans, our inability to deal with the
- 20:53Speaker 2: climate crisis is a lack of cooperation and digital intelligence.
- 20:59Speaker 2: They won't have egos, they won't be like clashing against
- 21:02Speaker 2: each other, so the saying goes, and so they won't
- 21:04Speaker 2: have any issue cooperating, and they also won't have any
- 21:09Speaker 2: lack of ideas and ability to experiment, develop new energy
- 21:13Speaker 2: storage solutions, develop new forms of renewable energy, and so
- 21:17Speaker 2: on and so forth. So that is the kind of
- 21:19Speaker 2: most charitable version of why artificial general intelligence could fundamentally
- 21:25Speaker 2: solve climate change. My critique is, again, we don't have
- 21:29Speaker 2: scientific evidence that AGI will ever come to pass, and
- 21:33Speaker 2: so we are basically trying to justify current day vast
- 21:39Speaker 2: environmental harms and the acceleration of the climate crisis with
- 21:43Speaker 2: a speculative possibility that it might one day be able
- 21:47Speaker 2: to go away. And so we're essentially trying to cover
- 21:51Speaker 2: up real scientific evidence of present day reality with a
- 21:57Speaker 2: spiritual belief that it'll all be okay in the end.
- 22:01Speaker 1: Listen. I would go further. And I mean my issue
- 22:04Speaker 1: with this is even if an AI agent, your chatbot,
- 22:07Speaker 1: can spit out the answer, Hey, here's what to do,
- 22:10Speaker 1: do this, do this, and do this, and climate change
- 22:13Speaker 1: will grind to a halt. What if we're not interested
- 22:16Speaker 1: in listening, you know what I mean?
- 22:18Speaker 3: Yeah? Right?
- 22:19Speaker 1: You could get on chat YOUBT, you can get on
- 22:21Speaker 1: claud you can get on Gemini or whatever and say, hey, hey,
- 22:24Speaker 1: AI friend, I'm I'm feeling really sick. I'm eating all
- 22:28Speaker 1: of this cake and I really don't feel good. What
- 22:30Speaker 1: should I do? And I'll say, hey, buddy, stop eating cake.
- 22:33Speaker 1: I'm gonna keep eating cake.
- 22:34Speaker 3: Yo.
- 22:34Speaker 1: You don't actually have to listen. The computer can't make
- 22:37Speaker 1: you listen, even if it has the answer. And here
- 22:40Speaker 1: in the States particularly, we've got the data. There are
- 22:44Speaker 1: some ideas on things we could do to reduce the
- 22:46Speaker 1: impact on the environment.
- 22:48Speaker 2: We're just not doing them exactly. I think this is
- 22:51Speaker 2: one of the things that has always broken down in
- 22:53Speaker 2: these theoretical future AGI arguments of whether it's going to
- 22:56Speaker 2: be fundamentally positively or negatively transformative. There's never a clear
- 23:01Speaker 2: articulation of how it's operationalized in the physical world. Are
- 23:05Speaker 2: they going to have robot bodies, are they going to
- 23:07Speaker 2: be mining the earth and developing and cultivating these new
- 23:12Speaker 2: energy storage solutions, or are they directing humans to do that,
- 23:18Speaker 2: at which point we still run into all of the
- 23:20Speaker 2: same problems that we've always right pad, which is humans
- 23:24Speaker 2: don't listen. So I think you're hitting upon one of
- 23:27Speaker 2: the core weaknesses of just many of the arguments in
- 23:32Speaker 2: this community is they do not actually acknowledge the social, political,
- 23:37Speaker 2: and economic aspects of the way that technology ultimately impacts society.
- 23:45Speaker 2: Like it's not just you have a technical capability and
- 23:48Speaker 2: everything is suddenly solved. There's so many more layers to
- 23:52Speaker 2: how the capability then translates into real world impact.
- 23:58Speaker 1: Yeah, I'm sympathetic to some of the ideas that are
- 24:01Speaker 1: put forth, like AI could cure diseases. Hey, look, you
- 24:05Speaker 1: can mix these three chemicals and it gives you a
- 24:07Speaker 1: pill and it's two bucks. Give it to everybody. Okay,
- 24:11Speaker 1: I'll buy that. And also healthcare obviously has a huge
- 24:14Speaker 1: social component to it. But yeah, something that truly is
- 24:18Speaker 1: linked to our behavior. I struggle to see how exactly
- 24:23Speaker 1: that's going to work.
- 24:24Speaker 2: What's interesting about the drug discovery or curing diseases thing, too,
- 24:29Speaker 2: is there's a word game that people within the AI
- 24:33Speaker 2: world play where they talk about when they talk about
- 24:37Speaker 2: AGA is going to cure cancer. What's confusing is that
- 24:41Speaker 2: there are plenty of AI technologies that can help us
- 24:44Speaker 2: advance and tackle that challenge, but it has nothing to
- 24:48Speaker 2: do with the type of AI that Open AI or
- 24:51Speaker 2: the rest of these Silicon Valley companies are building. So
- 24:53Speaker 2: they play this game where they just say AI and
- 24:57Speaker 2: AI is a huge umbrella term. It's like the word transportation,
- 25:00Speaker 2: like you could be talking about a bicycle or a bus,
- 25:04Speaker 2: or a gaskillsing truck or a rocket. These are all
- 25:06Speaker 2: different forms of transportation, and they're building rockets, but they're
- 25:12Speaker 2: pointing to the benefits of bicycles and public transport. And
- 25:16Speaker 2: so when these companies say AI is going to help
- 25:19Speaker 2: us cure diseases. There are plenty of AI technologies that
- 25:23Speaker 2: literally have no relation to what they do that are
- 25:26Speaker 2: making positive impacts on healthcare. There are machine learning models
- 25:30Speaker 2: that can be trained on MIRI scans to identify cancer,
- 25:34Speaker 2: and there have been studies that show that if you
- 25:35Speaker 2: give these tools to trained radiologists, they will be able
- 25:39Speaker 2: to identify cancer far earlier with a much higher accuracy,
- 25:43Speaker 2: such that patients can actually intervene early on and have
- 25:47Speaker 2: a much higher likelihood of kicking that disease. There is
- 25:50Speaker 2: also Last year, the twenty twenty four Nobel Prize in
- 25:54Speaker 2: Chemistry was awarded to a team at DeepMind that pre
- 25:58Speaker 2: them joining this large language model race. They developed this
- 26:01Speaker 2: tool called alpha fold and it was able to predict
- 26:04Speaker 2: with extremely high accuracy all of the protein structures of
- 26:08Speaker 2: different amino acids, and that in and of itself is
- 26:10Speaker 2: going to be one extremely critical building block for understanding
- 26:14Speaker 2: disease and for discovering new drugs. That was trained on
- 26:19Speaker 2: extremely clean, highly curated data sets. It was just amino
- 26:24Speaker 2: acids and protein folding structures. That was a very task
- 26:27Speaker 2: specific AI model that then was able to do incredible
- 26:32Speaker 2: things because fundamentally it was a well scoped highly computational problem,
- 26:36Speaker 2: and that's what AI is good at. You throw AI
- 26:39Speaker 2: at a highly computational problem and it will compute. But
- 26:42Speaker 2: what these companies are doing, and the critique that I
- 26:44Speaker 2: have is they pretend that they're trying to build everything machines,
- 26:48Speaker 2: and they're trying to do that by then scraping the
- 26:51Speaker 2: entirety of the English language Internet with just a boatload
- 26:54Speaker 2: of polluted data that has nothing to do with healthcare.
- 26:58Speaker 2: But it's just like people throwing curse words at each
- 27:00Speaker 2: other online and then they're like, this is going to
- 27:04Speaker 2: cure cancer, and it's like what are you smoking? Like, like,
- 27:11Speaker 2: how is that gonna get us to? Like what? Like,
- 27:15Speaker 2: we already have these other AI technologies that are making
- 27:19Speaker 2: these advancements, that are being trained on actual, high quality
- 27:23Speaker 2: data sets. And then you're gonna pump a bunch of
- 27:26Speaker 2: gunk into this large, nebulous, large language model that really
- 27:31Speaker 2: has very little articulated purpose and say it's going.
- 27:35Speaker 1: To do the same thing and pump gunk can do
- 27:37Speaker 1: the environment.
- 27:38Speaker 2: And pump gunk into the environment and exploit a lot
- 27:40Speaker 2: of labor and get like potentially automate away a ton
- 27:44Speaker 2: of jobs in the process.
- 27:46Speaker 1: So, yeah, there are people in Silicon Valley who love
- 27:48Speaker 1: to pitch AI as a silver bullet for everything from
- 27:51Speaker 1: cancer to climate change. But of course the reality is
- 27:55Speaker 1: not that simple. What this company or that company you're
- 27:58Speaker 1: building could really solve some of these problems. But as
- 28:02Speaker 1: they're doing it, they're also doing things that you might
- 28:05Speaker 1: have read about before, not in Forbes or in Wired magazine,
- 28:09Speaker 1: but in your middle school history textbook. Companies in the
- 28:12Speaker 1: industry are starting to act a lot more like empires,
- 28:16Speaker 1: Empires that extract resources, expand rapidly, and then leave communities
- 28:21Speaker 1: to deal with the consequences. We'll get into some of
- 28:24Speaker 1: those consequences after the break. Some of the consequences of
- 28:40Speaker 1: AI's colonialism you might already be able to imagine, like
- 28:44Speaker 1: when Google went to a part of Chile that was
- 28:46Speaker 1: in the middle of a huge water crisis and tried
- 28:49Speaker 1: to build a massive data center. If you think back
- 28:52Speaker 1: to our episode on AI and the Environment, you know
- 28:54Speaker 1: that data centers use up a ton of water that
- 28:58Speaker 1: didn't seem to matter to Google. But it's not just environmental.
- 29:03Speaker 1: In the book, Karen also talks about how the industry
- 29:06Speaker 1: always seems to go to countries undergoing some kind of
- 29:09Speaker 1: economic crisis and then find workers there to exploit. Karen
- 29:14Speaker 1: found that open ai used an intermediary company to contract
- 29:17Speaker 1: workers in Kenya for less than two dollars an hour
- 29:20Speaker 1: to build automated content moderation filters. For these filters to work,
- 29:24Speaker 1: you need humans to catalog, sometimes hundreds of thousands of
- 29:28Speaker 1: examples of things like the graphic content that OpenAI wanted
- 29:32Speaker 1: to prevent its model from generating. One worker on the
- 29:35Speaker 1: quote sexual content team had to review fifteen thousand pieces
- 29:39Speaker 1: of sexual content per month. And we're not talking just
- 29:43Speaker 1: regular porn, I also mean child sexual abuse material. Some
- 29:48Speaker 1: of this material was even generated by open AI's own software,
- 29:51Speaker 1: so that again they could check if the filters were
- 29:54Speaker 1: catching the bad stuff. But the workers who were manually
- 29:58Speaker 1: sifting through this stuff day after today, it caused some
- 30:01Speaker 1: serious mental consequences environmental exploitation, worker exploitation. This sounds like
- 30:08Speaker 1: straight up empire behavior, and it feels familiar. You could
- 30:12Speaker 1: take a map of the colonial powers and the colonies
- 30:15Speaker 1: like India, Latin America, parts of West Africa from a
- 30:18Speaker 1: couple centuries ago and lay it over a map of
- 30:21Speaker 1: where these AI companies are exploiting now, and it would
- 30:25Speaker 1: line up almost exactly. So we've read about this stuff
- 30:29Speaker 1: in the past in school. So now what this brings
- 30:32Speaker 1: me to a question? Though I don't know if this
- 30:33Speaker 1: is pushing back or maybe I'm just a little bit
- 30:37Speaker 1: pessimistic here, But if you talk about colonialism, you know,
- 30:40Speaker 1: empire's imperialism, those words mean different things in different places. Absolutely,
- 30:45Speaker 1: I think if you go to a place that was colonized,
- 30:47Speaker 1: it's going to bring back memories of slavery, members of exploitation,
- 30:52Speaker 1: memories of poverty, starvation, generations of wrecked governments, things like that.
- 30:57Speaker 1: For an American, we think colonial is a furniture style,
- 31:01Speaker 1: you know what I mean. We think it's a we
- 31:02Speaker 1: think it's a cool way to build your house. Oh God,
- 31:05Speaker 1: let's be real. That is so real. We'd be real.
- 31:08Speaker 1: And so I think about this, and I think about
- 31:11Speaker 1: your explaining imperialism of AI companies, and I think one
- 31:16Speaker 1: of the features of an empire, one of the features
- 31:18Speaker 1: of a colonialist empire is not only the leaders, but
- 31:23Speaker 1: the people who live there also start to believe that
- 31:26Speaker 1: that's just the natural way of things, that oh, there
- 31:28Speaker 1: are people in Kenya who are being forced to see
- 31:31Speaker 1: just absolutely horrific carnage imagery and getting paid cents and oh, well,
- 31:38Speaker 1: that's just the way things go. That hey, third world country,
- 31:42Speaker 1: Global South, that's what happens there and it doesn't happen here.
- 31:45Speaker 1: Let's just be real. How do you make a reader
- 31:48Speaker 1: who is in the empire understand that?
- 31:51Speaker 2: You know, I went on book tour. I stopped in Seattle,
- 31:55Speaker 2: and I had this wonderful opportunity. You talked to Ted Chang,
- 31:58Speaker 2: one of the most decorated science fiction writers ever, and
- 32:03Speaker 2: we were talking about this exact question, and he told me,
- 32:05Speaker 2: and I think he's exactly right. He was like, your
- 32:08Speaker 2: job is not to convince the people that are already
- 32:10Speaker 2: convinced of something completely ideologically opposite to you. If someone
- 32:16Speaker 2: already is convinced that there should be a hierarchy in
- 32:19Speaker 2: the world and that certain people don't deserve fundamental, basic
- 32:22Speaker 2: human rights like you do, not waste your time convincing them.
- 32:27Speaker 2: Who you're trying to convince is the broader public and
- 32:29Speaker 2: people who just don't really know how to think about
- 32:32Speaker 2: this technology and don't really know how to interface with it. Ultimately,
- 32:37Speaker 2: all empires are made to feel inevitable, but all empires
- 32:40Speaker 2: fall because the majority of people under subjugation end up
- 32:45Speaker 2: rising up and protesting the empire, and those are the
- 32:48Speaker 2: people that you should be speaking to. Another feature of
- 32:50Speaker 2: empire building is there are people who are richly rewarded
- 32:54Speaker 2: by empire, usually the people that are most powerful politically
- 32:58Speaker 2: and economically. Those are the people that are rewarded, and
- 33:00Speaker 2: that's why empires are able to perpetuate so long in
- 33:03Speaker 2: the first place, because the people who are totally coddled
- 33:06Speaker 2: into believing that this is a great state of affairs
- 33:09Speaker 2: are the people that then have the most access to
- 33:13Speaker 2: all of the levers to maintain the status quo. And
- 33:15Speaker 2: so he was like, don't talk to those people. Those
- 33:17Speaker 2: are not the people that you should focus on, because
- 33:20Speaker 2: they're never going to change their minds, like both philosophically
- 33:24Speaker 2: because they don't see a problem with an exploitaive, extractive worldview,
- 33:29Speaker 2: but also because all of the evidence that they are
- 33:32Speaker 2: exposed to reinforces the idea that things are just fine,
- 33:37Speaker 2: and so focus on everyone else, focus on rallying the
- 33:41Speaker 2: majority of the world around this idea. When I tell
- 33:45Speaker 2: people this empire metaphor, outside the US, no one has
- 33:49Speaker 2: ever questioned me. When I was talking with Chilean water
- 33:53Speaker 2: activists for my book about the expansion of data centers
- 33:57Speaker 2: in their community, they were first to bring up the
- 34:01Speaker 2: relation to their history, their colonial history. Really yeah, I
- 34:05Speaker 2: didn't actually even say it. They were like, what's happening
- 34:09Speaker 2: now is what's happened to us for centuries, first at
- 34:14Speaker 2: the hands of Spanish colonizers, then at the hands of
- 34:17Speaker 2: American multinationals, and now at the hands of new American multinationals.
- 34:21Speaker 2: And to your point that, like every different country experiences
- 34:25Speaker 2: colonialism differently, I mean, it was remarkable how they still
- 34:29Speaker 2: experience it differently, but in exactly the same way as before.
- 34:32Speaker 2: So in Kenya they were like, there is a connection
- 34:36Speaker 2: between slavery and labor exploitation happening now with the AI industry.
- 34:41Speaker 2: And in Chile they are a country that has been
- 34:45Speaker 2: extracted for their natural resources again and again. And the
- 34:48Speaker 2: term extractivism, which is an anti colonial term that refers
- 34:54Speaker 2: to the idea that massive amounts of resources are extracted
- 34:56Speaker 2: from one place and used to benefit a far away place,
- 35:00Speaker 2: no benefit goes to the local community. That was originally
- 35:03Speaker 2: about the Spanish and Portuguese term coined by scholars in
- 35:06Speaker 2: Latin America extractavismo or estrachivismo in Spanish and Portuguese, and
- 35:13Speaker 2: they were like, this is extractivism, right, We've seen this.
- 35:16Speaker 2: There's a connection between the Spanish colonial extractivism and the
- 35:21Speaker 2: air industry's extractivism. So it literally is a replaying. Yeah,
- 35:26Speaker 2: and people who live that history and understand that history,
- 35:29Speaker 2: there's no leap that they have to make to connect
- 35:32Speaker 2: the two.
- 35:32Speaker 1: That makes a lot of sense. And to be clear,
- 35:36Speaker 1: this isn't just happening in what we consider quote the
- 35:39Speaker 1: global South. It happens in the US too. That example
- 35:43Speaker 1: of Google trying to build a data center in Chile
- 35:45Speaker 1: despite the residents not wanting it might sound familiar. In
- 35:49Speaker 1: our episode on Ai in the Environment, we talked a
- 35:52Speaker 1: little bit about Memphis, where Elon Musk's company Xai has
- 35:56Speaker 1: been running gas turbines without permits for months now. It's
- 36:00Speaker 1: making people there sick. Just a couple of weeks ago,
- 36:03Speaker 1: More Perfect Union published a deeper dive on what's happening
- 36:06Speaker 1: in Memphis that gives some more details, and Xai has
- 36:10Speaker 1: already started building a second data center. So in Chile,
- 36:15Speaker 1: the local people organized and successfully stopped Google from building
- 36:19Speaker 1: that data center in Memphis. That fight is still ongoing.
- 36:23Speaker 1: But even if they're able to shut those turbines down tomorrow,
- 36:27Speaker 1: what happens to the people who were already hurt by
- 36:29Speaker 1: what XAI has done to the environment there. It's just
- 36:33Speaker 1: another example of what Karen Howe describes as the rise
- 36:36Speaker 1: of AI empires, taking resources, dodging accountability, and leaving communities
- 36:42Speaker 1: to deal with the consequences. You talk about some potential
- 36:46Speaker 1: alternative ways that this is going to be used or
- 36:48Speaker 1: this could be used, and so you're not necessarily talking
- 36:51Speaker 1: about everybody turn off the computer type thing. But yeah,
- 36:54Speaker 1: just tying in with the title, where's the kill switch
- 36:57Speaker 1: on this thing? And what hitting the kill switch?
- 37:00Speaker 2: So to me, hitting the kill switch in this context
- 37:03Speaker 2: is killing the imperial conception of AI development. Not killing
- 37:08Speaker 2: all AI development, but the imperial one where people at
- 37:11Speaker 2: the top can just say this is how it's going
- 37:12Speaker 2: to go and then consume the entire world of resources
- 37:16Speaker 2: in a pursuit of a Morphis vision of progress. The
- 37:19Speaker 2: thing that I want to see and I think how
- 37:21Speaker 2: we can get there. I want to see broadly beneficial,
- 37:25Speaker 2: task specific AI models that are developed and deployed through
- 37:30Speaker 2: the participation of communities. And when you think about the
- 37:35Speaker 2: AI supply chain, which I try to make visible in
- 37:37Speaker 2: my book, there's data, there's land, there's energy, there's water,
- 37:41Speaker 2: there's labor. There are spaces that companies need access to
- 37:45Speaker 2: deploy their technologies, like schools and hospitals and government agencies.
- 37:48Speaker 2: Silicon Valley has done a really great job over the
- 37:50Speaker 2: last ten years of making people feel like these resources
- 37:53Speaker 2: in these spaces are actually owned by Silicon Valley, but no,
- 37:58Speaker 2: they're owned by us. That day is your data, it's
- 38:01Speaker 2: my data. That in intellectual property is the intellectual property
- 38:04Speaker 2: of artists, writers, creators. The schools that's collectively owned by
- 38:08Speaker 2: teachers and students. Those hospitals are collectively owned by doctors, nurses,
- 38:12Speaker 2: and patients. And we're already seeing movements around the world
- 38:16Speaker 2: of people actually fighting back and reclaiming ownership of those
- 38:20Speaker 2: resources in those spaces. So artists and writers that are
- 38:22Speaker 2: suing these companies saying no, you can't just take our
- 38:24Speaker 2: intellectual property. The Chland water activists that are write about
- 38:27Speaker 2: in my book who said no, you can't just take
- 38:29Speaker 2: our fresh water and successfully stalled Google from building a
- 38:33Speaker 2: data center within their community for now five years, many
- 38:37Speaker 2: many movements around the world that are replicating that to
- 38:40Speaker 2: push back against data centers, teachers and students that are
- 38:42Speaker 2: having public debates now saying do we actually want AI
- 38:45Speaker 2: in our schools and if so, under what terms? Because
- 38:49Speaker 2: we wanted to foster curiosity and create a critical thinking,
- 38:53Speaker 2: not just totally eroded away. And if we can have
- 38:57Speaker 2: those conversations one hundred thousand times full and start moving
- 39:02Speaker 2: more towards task specific AI technologies, we will get to
- 39:07Speaker 2: a place where we do have AI that is broadly
- 39:11Speaker 2: beneficial and actually works for the people rather than us
- 39:15Speaker 2: working for AI.
- 39:17Speaker 1: Thank you, Thank you so much. Hope to be able
- 39:19Speaker 1: to chat with you again.
- 39:20Speaker 2: Thank you so much. Dexter.
- 39:26Speaker 1: All right, so this is the part of the episode
- 39:28Speaker 1: where I usually have some kind of closing thoughts, you know,
- 39:30Speaker 1: at my little two cents, three cents go on a
- 39:33Speaker 1: little bit, But this time I'm going to keep it
- 39:35Speaker 1: to four words. Go read Karen's book. Seriously, Just go
- 39:43Speaker 1: read Karen's book. One thing we didn't talk about, and
- 39:46Speaker 1: one thing I really do like about the book that
- 39:48Speaker 1: I should say is that you could pick it up
- 39:50Speaker 1: with absolutely no idea how AI works and you'll not
- 39:54Speaker 1: only understand the societal implications that you know we kind
- 39:57Speaker 1: of talked about in this episode, but you'll come up
- 40:00Speaker 1: with a better understanding of the technology of AI than
- 40:03Speaker 1: most people, even if you're one of those people who
- 40:06Speaker 1: says you don't like computers, you hate computers, you don't
- 40:08Speaker 1: understand them for real. It breaks it down in a
- 40:11Speaker 1: way that I've never seen done before. So highly recommend it.
- 40:14Speaker 1: And if you've already read the book or you just
- 40:16Speaker 1: want to talk about something else, let us know what
- 40:18Speaker 1: you think. We're on Instagram at kill switch pod, or
- 40:22Speaker 1: you can hit me at dex Digi that's d e
- 40:25Speaker 1: x DGI again on Instagram or am alsa Wan Blue Sky.
- 40:30Speaker 1: And if you haven't done it yet, leave us a
- 40:32Speaker 1: review on your favorite podcast platform. People actually read those things,
- 40:37Speaker 1: and your review could be the thing that convinces someone
- 40:39Speaker 1: or someone's to check us out. And this show is
- 40:43Speaker 1: hosted by Me Dexter Thomas. It's produced by Shena Ozaki,
- 40:47Speaker 1: Darluk Potts, and Kate Osbourne. Our theme song is by
- 40:51Speaker 1: me and Kyle Murdoch, and Kyle also mixed the show
- 40:55Speaker 1: from Kaleidoscope. Our executive producers are Ozma Lashin, Mangesha Gadur
- 41:00Speaker 1: and Kate Osborne from iHeart Our. Executive producers are Katrina
- 41:05Speaker 1: Norville and Nikki Etor catch On the Next One, Goodbye,