Tech Won't Save Us - The Threat of Data Colonialism w/ Ulises A. Mejias & Nick Couldry

Episode Date: October 24, 2024

Paris Marx is joined by Ulises A. Mejias and Nick Couldry to discuss how Silicon Valley's extractive data collection regime and the power it grants them resembles a much older form of exploitation: colonialism.Ulises A. Mejias is a professor of Communication Studies at SUNY Oswego and Nick Couldry is a professor of Media, Communications and Social Theory at the London School of Economics. They are the co-authors of Data Grab: The New Colonialism of Big Tech and How to Fight Back and among the co-founders of the network Tierra Común.Tech Won’t Save Us offers a critical perspective on tech, its worldview, and wider society with the goal of inspiring people to demand better tech and a better world. Support the show on Patreon.The podcast is made in partnership with The Nation. Production is by Eric Wickham. Transcripts are by Brigitte Pawliw-Fry.Also mentioned in this episode:Read an excerpt of Ulises and Nick’s book.Ulises has helped advance the Non-Aligned Technologies Movement.The World Economic Forum and Accenture published a report on governance of AI.Geoffrey Hinton was one of the winners of the Nobel Prize for Physics. Paris wrote about why we shouldn’t trust his assessment of AI.Google told the UK Labour government it will be left behind in the AI race if it doesn’t do what the company demands.Data centers use 21% of electricity in Ireland, and number that could jump to 31% within the next three years.Home building in West London could be restricted until 2035 because data centers have used up the available energy.Kenya is being drafted into the US’s anti-China tech alliance, which includes building data centers while ignoring the poor working conditions of data labelers and content moderators.Support the show

What's happening with data and AI is the seizing of the assets of the planet, but in a very particular form, which is the seizing of an asset that hadn't fully been captured before, which is human life. Hello and welcome to Tech Won't Save Us, made in partnership with The Nation magazine. I'm your host, Paris Marks, and this week my guests are Ulysses Mejias and Nick Coultery. Ulysses is a professor of communication studies at SUNY Oswego, and Nick is a professor of media, communications, and social theory at the London School of Economics. They're both the authors of Data Grab, The New Colonialism of Big Tech and How to Fight Back, and also of Costs of Connection. I had the great pleasure of meeting Ulysses and Nick back in May when we were both speaking at Republika in Berlin. And at the time, I had actually already read their

book, Data Grab. So, you know, it wasn't something that just came on my radar at that moment. And it was really great to meet them because I really enjoyed the book personally. I think that we see a lot of comparisons about particular historical frameworks that what's happening with the tech industry fit into. But when I read their explanation about data colonialism and how they see what is happening today as this extension of this colonial relationship, that really resonated with me, particularly the way that they described it throughout the book. And so I knew I wanted to have them on the show. It was just finding the right time with, you know, their schedules and my schedules in order to do it. So I was thrilled that we could finally set this up and discuss the book, discuss data colonialism, what it means and the wider implications of this concept for both how we think about technology, but also, you know, how we use it as users and what this way of developing digital technology and digital

platforms means for human society itself. So in this conversation, we explore many angles of that, right, from what it means on the platforms itself, how these platforms are constructed in this particular way to the more infrastructural angle of it as these things kind of actually become physical, are part of the land and are really transforming parts of our communities and the world itself to further entrench this very harmful and exploitative system that Ulysses and Nick are describing in this book. So as I said, I was really happy to have them both on the show. I think that you're really going to enjoy this conversation. I certainly enjoyed being able to dig into the book with them and ask them more questions

about data colonialism and its impacts. If you enjoy this conversation, make sure to leave a five-star review on your podcast platform of choice. You can also share the show on social media with any friends or colleagues who you think would learn from it. And if you do want to support the work that goes into making Tech Won't Save Us every single week, you know, from making projects like Data Vampires to having really in-depth conversations like this one you're about to listen to, you can join supporters like Michael in St. Louis and Sam in Mexico City by going to patreon.com slash techwontsaveus and becoming a supporter as well. Thanks so much and enjoy this week's conversation.

Ulysses, welcome to Tech Won't Save Us. Thank you very much, Paris. Happy to be here. And Nick, really happy to have you on the show. Yeah, it's great to be here. Thanks a lot, Paris. Awesome. So you both had this great book, Data Grab, which gets into this concept of data colonialism that I wanted to explore with you both. And so to get us started, I was wondering, how would you explain data colonialism to someone who has never encountered this concept before? Well, I think where we'd start is to say, look, we all agree a lot is happening with big data, big tech, and now AI. Something big is going on, and it's about taking data for sure.

Most people explain that through the framework of capitalism, and obviously it is part of capitalism, but is it something deeper? And in thinking about this, we go back to ask the question, what is the absolute core of colonialism? What is that thing without which there would have been no colonialism? There would have been no point in doing colonialism. And that is the land grab. The violence was terrible. The racism is terrible. But without the land grab providing the point for those, they wouldn't have happened. It might have happened anyway for other reasons, but they wouldn't have happened in exactly that way. They were the means to get the land, the land grab. And what we're saying is that not necessarily through the same means at all. We'll certainly

come back to that. There are big differences here. What's happening with data and AI is the seizing of the assets of the planet, but in a very particular form, which is the seizing of an asset that hadn't fully been captured before, which is human life, which for the past 20 years, it's been possible since we live a lot of our lives online to seize in the form of tracing what we do and then trapping that data and exploiting it. That's the core of what we're saying. And that's why we believe it's a really convincing analogy to the core of colonialism. Yeah. And I think, you know, since we're talking about our book, I do want to give our listeners sort of like the official definition that they will find in the book of data colonialism. Nick and I argue that data colonialism is an emerging social order, specifically for the appropriation of human life, as Nick was just saying, through this medium of

digital data, so that data can be continuously extracted for two reasons, mostly for the generation of wealth, for the generation of profit, and also for social control. So that's how we define data colonialism. And we're not against data. We're not saying data is bad. Obviously, we all generate it. We all rely on it for different things like checking the weather, you know, trying to figure out if it's going to rain today. So we're specifically talking about this use of data, this extraction, this appropriation of data for these specific purposes, generation of wealth, which also generates inequality and also for social control. And, you know, we do realize that using the word colonialism can

be tricky. Obviously, it's a word with a long history. Some people might, in fact, may perhaps even resent our use of it in the sense that we might be equating whatever's happening with historical processes that, as we know, were brutal, consisted in many cases of genocide, of extraction of resources. We are very much aware of these differences. In our work, we do acknowledge that historic colonialism is very different from data colonialism as we see it. There are differences in terms of the modes, the scales, the intensity. Colonialism in India was very different than colonialism in my native Mexico, very different than the kinds of colonialism we might be seeing today in Palestine. But we do argue that there is one crucial similarity across all of these different

forms of colonialism, and that is the function. The function is the same, and that function is to extract, to dispossess. So whereas before we were talking about territory and extraction and the appropriation of territory, today we're talking about the extraction and the appropriation of data. The modes and intensities might be different, but that function is still the same. Yeah, I appreciate you both outlining that. And, you know, like I've told you both, I read a lot of books about tech and a lot of books that make different comparisons. But I found that your comparison in reframing this around colonialism to be a really compelling one, right, for what we are seeing today and what we are experiencing

with this tech industry and everything that it is doing and the model that it has developed over the past number of decades in order to, you know, become so commercially successful, right? And so I wanted to dig down into this a bit more with you, right? Because you talk about that land grab and this progression at this point to wanting to seize the understanding of human life or, you know, whatever it is about human life, because we know that capitalism always has to commercialize more and more aspects of our life to continue growing and what have you, right? So you talked about this comparison to colonialism, how the land grab aspect of it and wanting to seize control of, you know, these aspects of human society and human existence is something that it has in common, but the violence is something that is quite different, right? And that we don't see so much with the expansion of these platforms and

with this model of data gathering and data control. So what do you see as the main similarities between, you know, say colonialism of the past and what you're describing here with data colonialism, but also the biggest differences that you see between the two and where it doesn't fit so well? Well, you're right to point out that violence is a big issue here, because we do mostly associate, when we think of colonialism, just brutal violence of that system. And so we want to be careful. We don't want to say there's a one-to-one correlation here. We're not trying to pretend that brutality of slavery, for instance, that happened during colonialism, we are seeing that in the exact same

form today. But we also don't want to pretend that violence is not part of this system because it is in different forms. So for instance, when we think about about algorithmic bias that we encounter, the way in which powerful algorithms and artificial intelligence has been found to discriminate against certain peoples, of course, then we continue to see a different kind of violence, again, propagated through the system. So it's important to keep in mind that there are important differences, but there are also certain continuities. And our work is about sort of excavating and exploring those continuities. The question of violence is a really fundamental one.

So I'm glad you brought that up early because we've talked about this to many, many audiences over the past five or six years. And it is one that for many people, they can't get over. They may have themselves experienced that violence in their family recently, or their whole history of their family is shaped by that violence. And of course, we take that seriously. That's one reason why we are so engaged in the history of colonialism. So why would there be a fundamental difference? Well, if we put it in simple terms, when the first colonization happened and through the early history of it, people were leaving European lands and finding human beings and territories. Sometimes they didn't even know

they existed. They certainly didn't know anything about those people and they probably didn't care. They had no social relations with those people. If you're trying to take something from someone you have no social relations with, there are only two options, brutal violence and lying. And they use both. Now, lying might still be going on in a subtle form. The brutal violence isn't quite so necessary. And there's a very interesting and fundamental reason for that. And it links back to this link between capitalism and colonialism. What's going on with data and AI today is to do with both capitalism and colonialism, the two together. In the book, we call it the double helix. You could never have had capitalism without the huge grabbing of resource from across the planet

that was colonialism. But where does that leave us today, two and a half centuries on? It means that we've spent two and a half centuries as human beings across the planet getting used to what capitalism demands of us, which is we sign terms and conditions. We know if we want that product, we've just got to accept the terms. There's no negotiation about that. That's a lot of violence being used in indirect means, because we're so used to capitalism. Just to give you a simple example of how there really is still something like violence going on, but it takes a radically different form. I was really shocked to find that marketers in the health data sector are tracking you just as much as the marketers in the market for sweets or trainers. Perhaps I don't care too much what they know about my body,

but I really do care that they know about my search data in relation to my life expectancy because that's of huge use to insurers. And this data is being traded. Those marketers are doing exactly the same thing. That's profoundly offensive and to many people, truly violent. And it's going on as if it was totally normal. So there are many forms of disguised violence, but the explicit violence just isn't necessary anymore because we're already plugged in to the system, which is a capitalist system, which started with colonialism. I appreciate you both outlining that. And, you know, I think bringing up that point of violence

is really important, right? As you were both talking about the examples that you were sharing, I was thinking also about things like predictive policing, right? And how we can very much see the use of this data in ways that are very harmful to people. And, you know, that's just one small example. There's so many more. But you talked about the history of colonialism there too. And before we continue to probe this concept of data colonialism, I wanted to return to that as well, because one of the points that you made in the book, which I found really compelling as well, was that, you know, when we think about colonialism, there was an aspect of Western technology that was very much connected to this colonial project, right, and enabling its growth around the world, but also, you know, the expansion of this capitalist

system that was driving so much of this colonialism as well. I wonder if you could expand on that a bit too, because I feel like, you know, when we think about what is happening today with digital technologies and whatnot, it can feel very disconnected from this larger discussion and this larger trajectory that we've been on for several centuries now. Yes, you're absolutely right, Paris. I think, as I mentioned earlier, in our work, we try to trace these continuities, you know, to use the past to help explain what's happening in the present. In order to do that, of course, we need to look at the history of colonialism to trace discontinuities.

You know, since we were just talking about violence, that I think it's important to keep in mind in terms of discontinuities is that a lot of what we're seeing today in terms of data colonialism continues to impact the traditional victims of colonialism disproportionately than the rest of us. So what do I mean by that? People of color, black people, women, and of course, the poor, you know, those three categories, I think, are essential that during colonialism, certain frameworks for rationalizing the oppression of these groups were put in place that continue up to today.

And that data colonialism, of course, also continues. We see this clearly not just in terms of algorithmic bias, but as Nick was just saying, if we start to think about the effects of health in terms of what's happening through the system of data colonialism, how it might impact the health of some people more than others. So yes, some of us might benefit in this system, but we do have to keep in mind that the traditional victims of colonialism continue to pay a heavier price. And you're right, the history of Western technology,

the history of Western science is very much a part of this discussion, because we do have to remember that Western science and Western technologies very much developed hand in hand, let's say, with the history of colonialism, which is not to say all of science is suspect. And we're not trying to argue that Western science is bad. Obviously, there are many benefits to it. There are many discoveries that have benefited humanity. So the point is not that. The point, however, is to acknowledge that a lot of Western science and technology was developed as a means to manage the colonized world.

The challenge, to put it in these terms, the managerial challenge was exactly that, how to manage colonies at a distance, when the center of empire was in Europe, Western Europe, and the colonies were in the rest of the world, what is now known as Latin America, or Africa, or Asia, etc. So this was a management problem. And so Western science and technology developed certain tools through the scientific method, through the application of science to technology, to develop tools of communication, to develop tools of transportation, to develop new tools in terms of military weaponry. All of it sort of geared towards making the management of these colonies possible.

And so I think that helps us reframe a lot of what we're seeing also in terms of these new technologies and artificial intelligence and algorithms, and also the power that some corporations are acquiring. Because let us also remember that colonialism was always a partnership between certain states and certain corporations. So it helps us sort of rethink a lot of these relationships and all of these new innovations in terms of these continuities. If I could just add a little footnote, continuing that to bring us maybe not right up to date, but 20 years in the past, rather than 500 years, what will be the equivalent today to these technologies of managing at a distance, as Ulysses said, which was the core of colonialism? I think it could be something as simple as the technology of Facebook. If you look at the early patents of Facebook, which I've been

doing recently, they literally lay out a technique in super technical language. And of course, using the whole architecture of patents, which is a key part of the scientific infrastructure to create a shadow world where you can't do anything without Facebook tracking it. It's literally impossible. There is nothing there except what Facebook has tracked. They had to build that world from nothing. So yes, this seems radically different from taking Brazil or taking a Caribbean island, because they pre-existed and they had their own utterly integrated history. Building through writing code and backing it in a patent, a territory where you can't do anything except that it's already captured is a very different thing.

But it has a broadly similar impact that you literally can't do anything there that what is managed. It is a way of managing you at a distance from a lab somewhere, even though you're the other side of the planet. And we feel very close, perhaps, to our social media accounts because it's our friends and our family on the other side. But in the middle is a mechanism for managing all of us at a distance. And you can see it laid out in something like the patent. And that's the creation of what we call in the book, a data territory, which is a brilliant concept to make the whole world treatable, just like like a territory because it's laid out right there in front of you and only you the engineer can see it the rest of us we're just inhabitants

of the territory i think that connects us really well to you know kind of the next stage of this right because we've talked a lot about this comparison to colonialism but now we want to get to the data side of this and i'll come back to data territory specifically in just a second i think what you were saying there nick really starts to get us close to this question that I want to ask. And that is, why is data so essential to the understanding of this concept, right? Obviously, we're talking about colonialism and this sort of relationship that comes with it. But why is data so important to both this analysis that you both are doing, but also this structure, this model that these companies have set up?

How does it give them so much power and so much control that makes this so concerning? That's a great question. And often people have said to us, well, surely what about digital? Digital colonialism, isn't that a better word? Problem is, when stuff became digital from being analog, nothing dramatic in itself happened. Yes, it was the start of something, but it wasn't the completion of something. We're all happy we now have digital data. If we didn't, we wouldn't be having this conversation now. So digital is not enough. Some people say, well, it's about

informational capitalism. Information has been flying around for the past three centuries of capitalism. It's not obvious there was such a fundamental break. But there was a fundamental break with the gathering of data about 30, 40 years ago, the beginnings of supply chains, the beginnings of credit card data. These were a radical change in the terms of business across the whole of capitalism. That's where we start to look for the origins. Why is it so important? Because data is a means to, first of all, track everything in a territory, and then secondly,

through tracking to control every possible parameter. And not just a few parameters, every imaginable parameter can be tracked and can be managed. So that creates a possible of much more integrated, multi-dimensional forms of management. Think of the Amazon warehouse floor. If there are still human workers there, and there are some, Every conceivable thing they do will be timed, measured, evaluated for its emotional input. If they have a sense of anger, it will be picked up instantly and tracked and compared with other forms of anger. This is what Marx imagined in terms of capitalist surveillance as management, but taken to a much higher dimension because there are no limits to the dimensions of data. That's why data brings about a fundamental change in the nature

of power relations and a fundamentally deeper and broader asymmetry of power relations between management who holds the data and the workers or the consumers who are held. I think it's important to sort of recognize that data is a form of abstraction. It abstracts our lives. It transforms them into something that can be quantified, that can be collected, that can be analyzed. So, you know, if we go back to Marx and his critique of capitalism, he was basically talking about how things that were outside of the market were brought inside of the market. And that's sort of the whole process of capitalism, how it functions, right? It commodifies things. And so when we looked at this modern phenomenon of digital data,

that's sort of the medium, the channel through which this happens in our lives. Because we live social lives, we interact with families, with friends, we go to work. None of that is necessarily part of a market. But what data did is to precisely transform all of these activities, all of this social richness, convert it into zeros and ones that could now be managed, that could now be analyzed and used even more recently to train large language models. So that's why data is very important, because without data, we wouldn't be able to abstract social life and bring it from outside of the market to inside of the market, where basically it can

be commodified and transformed into wealth. Hey, listeners, I'd like to take a quick moment to shout out another show I think you'd really enjoy. Cited is a podcast about the politics of science, research, and expertise. It's a reference to citation, you know, like an academic citation. Anyways, they have a new season that is definitely worth checking out. It's called The Use and Abuse of Economic Expertise. Economics sometimes feels like a physics, so sturdy, so objective, and so immutable. At least, that is the way that politicians tend to talk about economic research.

Yet behind every clean number or eye-popping graph, there is usually a rather messy story. A story shaped by values, interests, ideologies, and petty bureaucratic politics. In the use and abuse of economic expertise, Cited reveals the hidden stories behind the economic ideas that shape our world. You can find Cited wherever you find your podcasts, or at CitedPodcast.com. That's C-I-T-E-D Podcast.com. in particular, to talk about how this data is used in these ways that are particularly concerning when you think about it through this colonial lens. So I wonder if you could expand on these particular concepts and these particular uses of data and these particular ways of setting up these digital environments that we use in order to maximize the collection of data, but also the use of this data in these ways that you both find very concerning. So what is it about these two concepts that are so important to your broader understanding here? Well, a data territory is an abstract space created by writing code and software that is

so configured that data can be continuously extracted from it by whoever built and controls that territory. That sounds innocent, but it's a way of describing all the platforms we're on, all the apps, everything. And of course, there are good apps and there are good platforms. We're not saying in itself any data territory is intrinsically necessarily bad, but it creates the possibility of power. Because if you wrote the code, if you wrote the software, then you are in control of the tracking of everything. Because if you wrote the code, if you wrote the software, then you are in control of the tracking of everything that goes in there. And that means that you have the possibility of managing what people do in that territory, because you can send signals, you can nudge people, you can give them only five rather than 10 options. So there's no

option to dislike stuff on certain social media. There is always an option to like and so on. Simple things like that. Some of those decisions are good. Some of them are bad. But that's a data relation because we're basically in a modulated form of social interaction, which is very different from everyday life, where behind the scenes, there are certain menus which are limiting the sorts of things we can say. And that is absolutely not like our experience of social life normally, unless we're in a job interview, maybe. And even then, we sometimes take risks and we're not physically stopped from doing so. We may not get the job.

But in a software, there literally is nothing you can do outside the menu. So that is a highly constricted form of data relation, which is therefore the perfect means to impose a certain control form of power that doesn't exist in everyday life and never existed until we spend our life online. So that's why, and it links to the point about abstraction that Ulysses made, which is the other side of it, many things that would be part of our conversation simply can't be. And we're not in control of that not being there because they've been abstracted out. And that abstraction is normally there to benefit those in power. And of course, we're seeing this very much in the political situation at the moment. I heard someone just

yesterday describing two or three days ago at a street action, describing how everything they want to say about the humanitarian crisis in Palestine, they literally can't say on the social media platform of a certain sort. They literally cannot say it. through complex parameters, because the person said to me, I speak, but no one hears it, because it's not in people's feet, because the algorithm controls what it is, how I appear to you. These are radically different forms of social relating, therefore the different forms of power relations, different forms of data relation, and they create this territorial power, which has these profound wider impacts on the way whole areas of society are being structured today. And we'll come on to some examples in a minute, I'm sure.

Yeah, maybe just to exemplify some of these data territories, because of course, you know, we are all familiar with social media, with search engines, with using an AI chatbot. Those are apparent. But I think there are all sorts of data territories that are a little bit more hidden, sort of beneath the surface that we might not realize what is happening in these bounded spaces. So, of course, we have ag tech, agricultural technology. The production of food has become a very data intensified process where companies are using artificial intelligence to design thousands of variation of seeds that can grow in extreme environmental conditions because,

of course, corporations have seriously damaged our environment and now they're going to provide us the seeds that are able to survive in that destroyed environment. But we also have health tech that Nick was just mentioning as another data territory. We have ed tech, what's going on in schools in terms of all of the capturing of data, not just by schools, but schools are partnering with third parties who then conduct all of this appropriation of data

for their own benefits. We have all forms of bossware for tracking workers at home, in their offices, workplace, warehouses, tracking every single movement. We have a war tech as well, which we are seeing evidence of this across all of the war fronts that we see today in Ukraine or Gaza, Lebanon, using data basically to come up with targets of victims, in many cases, civilian victims. So each one of these functions as a data territory. There's a corporation or two, and probably a government or two behind them looking for new ways in which to use data to control that bounded space, again, for generating profit

or for controlling socially what goes on in that space. I think that makes perfect sense, right? And I find both of your descriptions there really compelling in thinking about the concerns that this presents, right? And I feel like on the one hand, you know, what you were describing, Nick, in relation to the limits that creating these platforms, these digital interfaces, you know, kind of bring into being, right? When things that we are used to doing in an old-fashioned traditional way then become replaced with some sort of digitized system that restricts what can actually go on here. I know

that speaking to people and reading about this for quite a while, that this is one of the concerns that comes with so much of the shifting of the interactions that we have, the way that we access services and all this, as it moves digital, it requires this kind of recrafting or reforming of how that service is actually delivered in ways that can really reduce people's ability to engage with it or, you know, to get the service that they actually need from it, because now all of a sudden your choices are limited to what works for however this system is designed. And then I feel like the broader piece of this, you know, when you're thinking about data, I feel like so much of the discussions that we have about how to respond to all this data and the power that comes from this data is always around, do we give people more rights to control

their data? Or are we going to pay people for their data or something like this? And I feel like never part of that discussion is, do we need so much of this data in the first place? Should we be creating all this data? But that doesn't really work for the broader conversation. So you don't hear it so much. Yeah. I mean, we see basically generative AI as a form of prospecting. We can talk about it sort of in terms of the tragedy of the commons, what happened in the early days of the internet. We were all led to believe that we were contributing to these wonderful commons where all of the data we were uploading, we were

sharing, would enrich humanity because we were all going to benefit from it equally. And then, of course, what happened is that these commons got enclosed. The boundaries were defined by corporations. And one day we woke up and we realized this wonderful community we were building online, now all of that data that we contributed belongs to a corporation. And they're putting real limits and applying this data in very specific ways, obviously, to create this wealth. By now, of course, with AI, we've gone from tragedy of the commons to perhaps a farce of the commons, because now the next step is now that we have enclosed the commons, let's take all of this data to create a machine that basically the whole point is that it's going to replace your job. It's going to replace even your creative outputs and the ability to make a living by making music, making art. Now we will need all of that because a large language model has been trained with all of the data that we have been producing for the last

couple of decades. And this machine is going to be able to replace human beings in the production of culture. So that's kind of a scary development in all of this. And if we just take that one stage further to, well, how is this even possible? Some of this even goes beyond our model of the restricted data territory in a way, because we're now finding out, as Ulysses was just saying, that even when we were making a song and it just happens to be online or putting our pictures up there or chatting to people or maybe just doing a little blog about something, all of that is fuel for AI. It can just be taken. And in fact, it needs to be taken. And we were very intrigued to see OpenAI when they gave testimony to the UK House of Lords. They actually came clean. And Sam Altman said in his written testimony, well, our business model depends on taking copyrighted material. Hang on a minute. Now, if you or I take copyrighted material, that's a crime. We're going to go to jail.

We're in big trouble. But when OpenAI with a smile take everything anyone has ever made, that's somehow not a crime. In fact, it's a necessity. And in fact, just today, by a weird coincidence, the World Economic Forum, which for more than a decade has been a cheerleader for everything data colonial, published a report called Governance in the Age of Generative AI, co-authored by someone from Accenture and an academic or someone from WEF. Well, they actually

cheerleaders for this very idea. They know that countries like Singapore, Estonia, and a few others have actually made exceptions to their own copyright law and say, no, it's totally fine to take everything you want, because that's either fair use, which is okay, always has been under copyright law, or we just make a big exception for generative AI in any event. And the words used, and this is an example of the sorts of, we call them in the book, civilizing stories that get told to sweeten the pill or repackage data colonialism, just as historical colonialism was repackaged. The idea is, well, that's necessary for innovation to advance science back to science again, or to create business value. And that must be all right, even though it breaks all the rules. As the New York Times is protesting in the courts at the moment, it breaks all the rules of capitalism, but somehow it's okay. So this is an example, and it's a very interesting one in the past two years, where in a sugared way, and open AI is, ChatGP is fun to use. It's a nice little game. Something, a land grab,

a data grab is going on, but it has to be disguised. And that's where you need the civilizational story that colonialism has been incredibly successful in generating for 500 years. We just have new versions of that same old story. We're taking it from you, but hang on, it's all going to be better. So just trust us. It's all going to be for the progress and betterment of humanity. And we were told that story with colonialism, and we're still being told that story today. In the book, we sort of talk about this narrative of the myth of cheap nature. So in order to colonize the world, of course, colonizers had to imagine the world,

the nature, territory as something cheap. It was abundant. It was just there for the taking. So colonizers could arrive in Australia, for instance, and say, you know, this is terra nullius, no man's land. It doesn't belong to anybody. We can take it. Of course, to transform that territory into wealth, you needed also cheap labor. So you go from cheap nature to cheap labor. And of course, colonialism was a racialized system. So the labor of black people, the labor of people of color, the labor of women was seen as, again, cheap, abundant, just there for the taking. We can just take that labor and use it to create more wealth. So let's follow that progression. We're going from cheap nature to cheap labor. And now,

of course, we have cheap data. That's the latest civilization or story. Cheap data has many of the same characteristics of cheap nature and cheap labor. It is, again, said to be abundant. It is free. And it is just there for the taking. Open AI can say, well, look at all of this data. It doesn't really belong to anybody. Yes, individuals produce it, but by the time it gets collected, we, companies, corporations, need to provide the advanced technology to refine it, just like oil refineries of the past, sugar refineries of the past during colonial times. So now corporations provide this advanced technology to take this refuse, this exhaust, and transform it into something useful. The story goes, the colonial narrative tells us,

again, this is happening, but the good news, the story says, it's for your benefit. It is for the betterment of humanity. Those narratives are so important, right? And the tech industry has shown itself to be so successful at creating them time and again, and perpetuating them through society. Thinking back to the things that you were saying, right, about the internet and this vision that we were sold of the internet as a commons and then how it was commercialized. But these ideas that came out of that era were still weaponized and used by these tech companies to get this kind of broader acceptance. And, you know, you've explained a lot. Both of you have explained a lot about the problems that this data colonialism presents us, right?

And the very real harms that come of, you know, this particular model that these tech companies have created. I think going beyond those civilizing narratives, what do you see as the way that these companies so successfully sell this to the public to not get this wide swath of the public saying, we're not okay with this, we don't want this to happen, and we want some other way of arranging society or the digital world or what have you? Well, I mean, we could take ChatGBT as an example. And it's a really interesting one because again, the news has been good today. We're speaking on the day when Geoffrey Hinton, one of the leading developers of neural nets, got the Nobel Peace Prize in physics. That, I have to admit, took my breath away. That was astonishing. Let's just step back for a moment. Yes, Geoffrey Hinton, without question, is one of the most thoughtful and skillful people in the area of data points, can be processed in a regularized,

automated way more effectively than it could be before. That could cure cancer. It could find the structure of proteins. No one's denied that that is positive in a good scientific context. But he's also the person who came to fame with the ImageNet data set, which he analyzed in 2012, which was actually the basis of the labels that human beings had attached. They'd been repointed through Amazon Mechanical Turk, which is another one of these human labor forms, which is forgotten in the history of artificial intelligence. There's a lot that's not artificial at all. And of course, it tends to be distributed in the global south, where people are more open to exploitation, or parts of the US where people can't get by and they need money through

Amazon Mechanical Turk. So his history is quite interesting. But then if you take ChatGPT in itself, there was a point in the development of neural nets when the most sophisticated people, the people working with Geoffrey Hinton, realized we've hit a brick wall because a machine is just a machine. It doesn't live in the world. It doesn't know really what a car is, even if we have hundreds of people to tell it. And the one thing it doesn't know is what is Nick and Ulysses Paris, what are they talking about right now? Why are they talking? How does this relate to what they just said?

It doesn't know context and human communication is based in context. So the great idea was to access context. How do you do that? Well, you create predictive machines that are sufficiently credible that human beings think it might make sense to interact with them by asking them a question and getting a reasonably credible answer, which is what we do with ChatGBT. And then if we like the answer, it's an answer. If we don't, then we feed that back to ChatGBT and it corrects the model. And next time it gives a slightly better prediction. Vin Cerf, who is one of the founders of the internet, came up with an incredible analogy for this. He called it a salad shooter. You just chop everything up into the maximum number of tiny bits, put it in a massive salad washer. It comes out as long as it's grammatically credible

as something like an answer to the question you just asked, that counts as the truth. So again, this is a weird form of serializing that makes something that is merely a mathematical prediction, okay, based in smart math, into a truth that rely upon sufficiently so that the person who devised it gets the Nobel Prize in physics. I mean, that is a question, right? Why don't people refuse? I actually think most people, once they understand some of the issues and what's involved, they do refuse. But as with the history of colonialism, you know, when people resisted colonialism, it rarely made the news until anti-colonial movements, movements of decolonization got big enough, important enough that people in the center

of empire had to pay attention. So, you know, in some ways that brings us to the point about resistance. I think people are resisting today. I think if you listen to your podcast, Paris, we get to hear a lot of those stories about resistance happening. Now, maybe it doesn't make it to mainstream news. Maybe it will in the future. But I do think, especially with generative AI, I think people are realizing some people are embracing it. I just look around my campus and I think it's a microcosm. Some people are embracing it.

Some people want to incorporate it in the classroom. And others, including students, are realizing, no, I'm not okay with this. In some ways, I think we need like a student bill of rights for the age of generative AI, because some students are saying, I won't use it in my assignments, but can you, the university, also ensure that you won't use it for grading? Professors won't use it for developing their lecture materials. Can the university ensure that you won't use my data that you collect when I visit the website or interact with the learning management systems? Can you ensure that you won't use my data to just feed an AI machine if I am agreeing that I won't use it to cheat on my assignments.

So I think we're coming to that point where more and more people will start to resist and to basically refuse some of these narratives that so far a lot of us have been believing in. Yeah. And I feel like even hearing what you're saying, I'm saying to myself, I need to do an episode on ed tech and the effect that this is having in education. But that's for another conversation. I want to talk to you a bit more about resistance. But first, I want to go back to one other point that you've both been making, right? You know, when we think about these generative AI models, and Nick, you were mentioning in particular, these governments around the world and these tech companies who are demanding that their collection,

their mass collection of data in order to feed these models and train these models be something that they're allowed to do, right? And it brought to mind this story in the UK recently, where Google basically said to the new labor government, you are going to be left behind in the generative AI race if you don't pass a bill that says that our taking and stealing of all this data to train our models is legitimate. But also there was another point to that where they said, and also if you're not building enough data centers to be able to properly compete in this. And, you know, we've been talking about data colonialism and data territories and things like that, but you mentioned how the original colonialism was very much linked to land, right? And grabbing land.

And I feel like there is a very physical aspect to what you're talking about as well, when we think about how these tech companies have been building out these major underwater fiber optic lines and increasingly trying to control those. And of course, the push that we're seeing now to build out these major data centers around the world. I'm wondering what you see as the infrastructural piece of data colonialism as well and that angle of it. I think in many ways, what we're trying to do in our work is to trace a line between the plantation, the factory, and the data center as three discrete moments in the history of colonialism and capitalism. And you're right, Paris, that it might seem like data colonialism is part of the cloud,

is something that happens in this ethereal space. But as you're pointing out, this is very much a conversation also about land and territory. And data centers make that incredibly apparent. So just like plantations were about land, appropriating land, and then developing monocultures so that countries would only be dedicated to producing sugar or minerals or whatever. And then factories also took the land and created these infrastructures for exploiting labor in new ways. We have data centers which occupy space, actual space in our universe, and use a lot of resources that are becoming more and more scarce. So as you have talked about in your podcast again and again, energy use, water use, the fact that we're building data centers in parts of the world where some of these resources are already scarce. And basically, we will need to take those resources away from people who need them to be able to feed these hyper data centers that are very hungry monsters that consume these

resources in the large scale. All of it while not paying taxes, all of it while being extremely secretive about what's happening in some of these areas, all of it while making special deals with governments that many times, most of the time are not, you know, advantageous to those government citizens. So I think that's plenty of reasons to start to, you know, to look at this also as a territorial issue. And, you know, the latest signs from, say, the UK government that you mentioned are not encouraging. Google already had a very close relationship with the British government through the NHS, along with Palantir. And Google was involved in various scandals through the use of NHS data,

which you probably heard about over in North America, and no doubt lay behind the UK government's 2023 pro-innovation, that was a phrase, regulation of tech, which is enthusiastically cited by the World Economic Forum just today, because it's exactly the direction they think everyone should be going in. But on the question of the data centres, one was recently opened or was proposed up in the northeast of England by the new Labour government. I drew breath at that point to think, hang on, haven't you listened to Paris Marx? Haven't you listened to the blogger? Don't you know the danger? Just to give you one homely example, I used to live in West London, which is a relatively prosperous part of London, but it needs houses for sure. And it's very ethnically mixed as close to Heathrow.

There will be no new houses being planned in that area until 2035. Why? Well, because the future supply of electricity, and these things are mapped out in big blocks by the big tech, has already been taken. Who by? Data centers, as reported by the Financial Times. Data centers. One third of Ireland's electricity is going to data centers. Now, you might say first world problem, but actually when that first world problem, and it's a problem, gets pushed down to the global south, which is even less chance of negotiating the terms of trade, you can be sure that when Kenya is told that for its future, it absolutely has to get the number of data centers

the World Bank thinks is consistent with a liberal market economy, there's going to be a lot of problems. In addition to the lack of luxury housing being bought, there's going to be many other costs for water, electricity, and so on. These are problems that will grow. It's not sustainable going in this direction, basically. And no one seems to be saying that, least of all governments. Yeah. The example that you gave in West London is such a shocking one. And you can give so many more examples of that for so many communities around the world, right? Which is where it really becomes just staggering. And I'm happy that you brought up

Kenya there because that was another thing I wanted to ask you both. And we've been talking generally about data colonialism, but do you see a distinction in how this plays out, say, in the global north and in the global south in the broader impacts of these things, you know, and the way that these models and these methods of, you know, data collection and using this data actually play out in different parts of the world? Well, I think there certainly are going to be differences, partly because of the legacy of historic colonialism, which we insist in our book has not gone away. Some people misread our book and say, well, we're saying data colonialism replaces historic colonialism. We're saying the exact opposite. Of course, historic colonialism has not gone away. We have a profoundly unequal

global economy. Racism disfigures every society on the planet linked to historical colonialism. And that means that the field in which data colonialism, the new toolkit of colonialism today will be tried out and then applied and then forced on people is a profoundly unequal. I'm not without hope. I think there's some very interesting things going on in Africa. The African Commission has a major consultation on AI legislation at the moment, which I hope will come up with something radical. Similar radical things have been happening in parts of Latin America. But we shouldn't deny the fact that these are operating against a very difficult situation. The supply of capital, the supply of development capital, the fact that these are operating against a very difficult situation. The supply of capital,

the supply of development capital, the fact that all the apps and the texts that are being used are already existing in the West or are in China, which in our book is the other pole of the new data colonialism. And that's another thing about today's world, which is radically more complex than the world of historical colonialism, when it was just Spain and Portugal initially, that means that it's going to be incredibly difficult to resist this tremendous push, which is so dominant, even in super rich countries like Canada, Ireland, Britain, Germany, and so on. Yes, I think, you know, obviously one thing that colonialism did was to divide the world into metropolises and colonies. And so since then, that distinction has been very real. Today, we speak mostly in terms of global north and global south. And it's true that each one has its own peculiarities and specific problems. So I think we need to acknowledge that, that data colonialism will look, will take very

different forms depending on the part of the world that we're talking about. Having said that, I do think that the other thing that colonialism did that we should keep in mind is that those boundaries also became very porous in the sense that as soon as colonialism started, we also have migration from the colonies to the center of the empires. And so the South became a part of the North and the North became a part of the South. And so we see that also replicated in data colonialism. Obviously, if we're talking about the gig economy in the global north, we can see how exploitation and extraction happens right in the middle south that basically replicate the same models, the platforms, the artificial intelligence models. They're copying them from the global north and applying them to the global south. So again, we have sort of miniature enclaves of the north

also in the south. So these relationships are very complex and very dynamic, and we can expect to continue to see those exchanges in data colonialism as well. Yeah, that makes perfect sense to me. And as a final question, you were talking a bit about resistance earlier, right? And how people might want to push back on this. And I feel like, you know, you also talked about how one pole of this data colonialism that you talk about is the United States. And I think, you know, this is the one that we would be most familiar with and the platforms that we most often interact with. But obviously, we see, you know, the Chinese model adopting a lot of the approaches data sovereignty or digital sovereignty becoming more common as more and more countries seem to want to get control of their digital infrastructures and the digital technologies that their societies use. And in the past, that seemed more focused on what China was doing and the way that it was carving itself out from the United States. But now it increasingly seems like something that is saying we reject what China is doing and what the United States is doing, and we're trying to form something else. So I wonder

if you see any hope in movements around the world or approaches around the world that are trying to push back against this Dana colonialism model, what you see is most hopeful there. For sure. This concept of sovereignty is very interesting because obviously when it means autonomy, when it means independence, I think that's definitely what we want. That's why some of us are sort of interested in this non-aligned technologies movement to replicate the original non-aligned movement, which some of your listeners might be able to identify an agreement between nations in the Cold War, who basically said, we don't want capitalism and we don't want communism. We need this third way. We need this autonomy and this independence.

Today, obviously, it's a little bit more complicated because Europe, for instance, wants to be sovereign because it doesn't want to depend on the United States or China too much. Of course, they're finding out it's impossible not to do that. But this notion, this definition of sovereignty is always problematic when it comes to the state, because it always includes certain people, but it also excludes others. Sovereignty, even in the time of independence, when colonies got their independence, was used precisely in this way to say who is in and who is out. No surprises, those who were usually out were indigenous people, people of color, people who were considered not good enough to be part of the nation state.

Today, when it comes to Europe, for instance, as it talks about data sovereignty, it is still using data to exclude certain people, migrants, for instance, or when it comes to European or companies from the global north developing tools of war, advanced and new tools of war. Obviously, the people who are out in that case are civilians in these conflict zones who might become part of a list of targets without regards for their civilian status. So I think this concept of sovereignty, we need to be careful. It's promising, but we just need to be aware that traditionally, historically, it's been used to keep certain kinds of peoples out. Just one final point, building what the hope that Ulysses has just talked about, about more resistant uses of data. Why not think about community sovereignty? Why does this always have to be

at the level of the state? In a way, the core of what we're saying is that this is not going to be easy to resist this. This is a massive thing, as large as earlier colonialism, as much as one key aspect of capitalism. So individuals can't fight back. We can only fight together. That's the only way that resistance even begins to make sense. But on what terms? The things that Ulysses is an expert about with sovereignty at the level of states are particularly difficult area and few people are very optimistic about that today. But we know there are wonderful experiments going on in Latin America and Africa and the Middle East and different parts of the

world where communities are saying, you know what, we've had enough. We're just going to imagine a future without that way of doing things and on a different basis. And actually, we have just about enough tech to make that happen. And we, as we said at the beginning, we're not in any way against data. We're in favor of data collectives which track male violence against women. We're in favor of data collectives that track the violence of sovereign states against migrant populations, or track its use in warfare, and so on. We're absolutely in favor of that. What we need, and this is the reason we wrote the book, is to flip things over and say, hang on, why wouldn't we start from what communities want? Tim McGebrow had a beautiful

phrase like that. I can't remember the exact wording. Why wouldn't we start with what the people who are being harmed by data want, rather than what the people who are doing the harm want? Why can't we shift the terms of the debate? That's why we wrote the book. And we hope people more and more will start to listen to that and take it into their own hands. It's such a great point and such a great place to leave our conversation as well. You know, as I said earlier, I would highly recommend people pick up the book Data Grab. I really enjoyed it. Ulysses, Nick, thank you so much for taking the time to speak with me. I really appreciate it. Thank you, Paris. Great pleasure. Thanks a lot.

Ulysses Mejias is a professor at SUNY Oswego, and Nick Coltry is a professor at the London School of Economics, and they are the co-authors of Data Grab. Tech Won't Save Us is made in partnership with The Nation magazine and is hosted by me, Paris Marks. Production is by Eric Wickham, and transcripts are by Bridget Palou-Fry. Tech Won't Save Us relies on the support of listeners like you to keep providing critical perspectives on the tech industry. You can join hundreds of other supporters by going to patreon.com slash tech won't save us and making a pledge of your own. Thanks for listening. Make sure to come back next week. Thank you.