Interview Transcript
Interviewer: Tom Grundy
Guest: Karen Hao
TG:
Welcome to a special episode of HKFP Yum Cha with the Karen Hao, a multi-award winning journalist working on AI, author of the best-selling Empire of AI (now out in paperback) and co-host of the BBC’s The Interface podcast. Karen studied mechanical engineering at MIT and was previously with the Wall Street Journal here in Hong Kong. This year she launched the AI Resist List to document examples of resistance to the AI revolution/apocalypse around the world. And she herself remains a rather lonely but bleeding critical voice on the bleeding edge of the Zeitgeist—all at a time when we’re awash with AI evangelists espousing the merits of a technology that’s disrupting business, politics, and the environment.
Karen, thank you for coming in to our humble headquarters in this new trillionaire era and for braving the rain. If you hear thunder, it’s because you’ve angered the AI gods, okay?
Karen:
Exactly.
TG:
I’ll let you just talk about the colonialism metaphor you rely on in the book.
Karen:
First of all, thank you for having me. I love being in a news office again.
The reason why my book is called Empire of AI is because I called the companies like Open AI "Empires of AI". "Empire" is the only metaphor that I’ve really ever found to encompass every facet of how these companies operate and how we should ultimately think about their role in society.
So, first and foremost, they amass an extraordinary amount of economic and political power and you could argue that they’ve become the preeminent or dominant power in today’s world, but they do that through the dispossession of the majority.
In my book, I talk about four specific parallels, but there’s honestly many more.
The first is that they like to claim resources that are not their own: the data of individuals, the intellectual property of artists, journalists, writers, etc.
The second feature is that they exploit extraordinary amount of labour. The workers that they use to produce their technologies see very little value in return; the workers are displaced through the deployment of this technology; ultimately they then accrue the value of the salary that would have gone to that worker because the company is instead buying their product or service.
The third parallel is that empires always control the flow of information in society, and the empires of AI do that both by censoring and controlling what kind of research is produced on AI. So the public is in the dark about the true limitations and capabilities of these technologies. Also these companies are ultimately producing an information technology where they want to be the portal through which everyone around the world uses to understand the rest of the world.
The final one is that they wrap this all up in a civilizing mission, saying that they’re doing this for the progress of all of humanity.
TG:
Right. Sam Altman, in the early days in particular, said, "It’s going to solve cancer, the housing crisis, the climate crisis, democracy, poverty, mental health, inequality." But they’ve also warned of deadly bioweapons, cyber security chaos, massive job losses, misinformation, literal existential risk. Sometimes they will have a model and they’ll warn, “Oh no, I’ve accidentally on purpose created something really dangerous. Again, somebody stop me."
Why is it important that tech chiefs seem to have to play to both hopes and fears?
Karen:
This goes back to the fourth parallel of why I I call these companies empires. It's because religion is a really powerful feature of empire building.
There is the civilizing mission, and that promises to bring humanity to a heaven-like state, but complementary to that, and what goes hand-in-hand with that, is the threat of descending into a hellish state. These are like the oldest stories that we’ve told ourselves throughout human history. It's the idea of utopia and dystopia, heaven and hell. By conveying that there are such extraordinary stakes, what these tech chiefs do is, they create this narrative of—or justification for—why they must retain total control over this technology’s development: If it falls into the wrong hands, then we’re doomed, but we could in fact access that utopia.
So, it doesn’t really work unless there is this kind of evil empire competing against what they are ultimately trying to achieve.
TG:
Do you feel they’re giving up more on the hopeful stuff lately? I think, you know, as money seems to coming more of a thing, they’re leaning on more on "it’ll reduce your head count and cut costs".
Karen:
I don’t think they’re leaning on it more than usual. I think there are just a couple dynamics happening.
One dynamic is that, within the AI world, different executives will lean on different dimensions of this narrative. Sam Altman is more traditionally leaning on the side of utopia and using the carrot to try and get people to go in this particular direction and allow his company to do whatever they want, whereas Dario Amodei of Anthropic typically leans more on the doomer narrative.
What’s happened in the recent months is that Anthropic has become much more of a dominant force and is beginning to out compete OpenAI both in terms of commercial value and also in narrative swaying power. We’re now hearing more of the doom narrative because that’s what Anthropic foregrounds in its storytelling, but both crutches or both legs of the stool have always been there from the very beginning of this industry.
TG:
Talking to "doom", this may be your first interview since the world burst its first trillionaire last week.
If we have this "underclass" created through massive unemployment, innovation being stifled by IP theft, and if the environment is ruined, what market is left to sell to? What even is the end game?
You went to school with some of these tech bros. What at this stage is their motive? What’s the point in a trillionaire?
Karen:
Um... I couldn’t tell you, haha...
TG:
If it’s money (maybe) in the early days, is it now power?
Karen:
I think money has always been a lever of power. There’s no point in accruing money if we didn’t live in a capitalistic system, where money brings you a significant amount of power. So, it’s always been about power.
We have reached a stage of capitalism where the system has allowed for the creation of these individuals with extraordinary extraordinary degrees of power as manifested in many ways, not just monetary, right? Elon Musk before he was a trillionaire already had an an exorbitant amount of political power and other forms of power, in addition to financial power.
So, I think that this is the perfect example of runaway capitalism and also imperialism, in a way that part of why I think these companies and individuals at their heads are so fixated on this idea of wanting ever more is because of this drive to capture everything under the sun, the way that an empire used to.
I mean, even though Elon Musk is the first trillionaire, you see this ideology parodied and manifesting among all of the the heads of these companies. Sam Altman, before he became the CEO of OpenAI, was the president of Y Combinator, a startup accelerator in the Valley, and he said back then, "I want to invest in 10x more companies year after year until I’m investing in every single company in the world."
It’s like there’s this drive to want to lay claim to everything. Until that happens, it’s not enough.
TG:
To run with the metaphor of "Empire", we saw this decolonization process after World War II, where the colonial powers were overstretched, the money was running out, and there were indigenous resistance movements coming up, there was this UN framework for states to sign up to, defining things like self-determination. With AI now, we’re seeing localized uprisings. The AI firms are looking overstretched in recent months. Last week, we HKFP signed up to join the SPUR coalition, which is a journalistic effort to have some framework for AI firms to sign up to, which can monetize the AI telemetry and try to correct some of the copyright issues.
Are there parallels or lessons when it comes to how to take down "Empire"?
Karen:
I absolutely think there’s parallels. I mean, one of the things we should take as a lesson from history is that empires are not inevitable. Even though they feel that way, every single empire has fallen in history. Part of it has been because there is a broad coalition of resistance movements that build up and then topple the empire.
I think the analogy holds in a way that what we are seeing now is that in the global conversation around AI, there’s been a significant change in sentiment. I mean, when I was working on the book just two years ago, the hype was out of control and everyone was parroting the industry’s narratives that this is going to be the best thing since sliced bread. Now what we’re seeing is many more people aware of the environmental harms, the labor exploitation, the public health harms, and questioning whether or not this is really the way that we should be approaching AI development, being deeply concerned about the impact on their kids and their critical thinking and their future economic opportunities and so on and so forth.
In the US, what we’re seeing is that 80% of Americans are now concerned about AI and it’s becoming a top issue in the midterm elections. Through this broad base of dissatisfaction, we are then seeing all these different resistance movements popping up and, in uncoordinated ways, independently beginning to push back and actually hold the industry accountable. It is having an effect on the ways that these companies are actually able to operate.
So I think what we saw with how empires of old ended up meeting their end is beginning to play out with the empires of AI.
TG:
Is it mostly environmental stuff that’s driving the push back?
Karen:
No, actually, every single harm that is coming out of these companies, I think, are all intersecting motivations for push-back.
I mean, environmental stuff is one of them, but labour is a huge one. When you have an industry saying "we’re coming after all of your jobs", that really is a huge mobilising and rallying call for a lot of people to think "wait a minute then, why are we building a technology to do that".
Psychological harm towards kids is a really big one, of course. That was a really big one for holding social media accountable as well. It's "why are we okay with platforms eroding the mental health of our teens". As people have realized that AI chatbots can engage in a similar kind of psychological impact, that is yet another thing that is galvanizing parents to be really critical of these companies.
The reason why there’s such a broad coalition of resistance happening is because there are so many different reasons why people are resisting and it is essentially touching every single asset and every single community in society now in these negative ways.
TG:
You said when you finished the book, it was all rather depressing, but this has totally changed over the last year or two. And now you’ve put together this AI Resist List, where I was looking through different examples from around the world, one in China even. But looking at the US, you now have push-back against the push-back. State-level national agencies are starting to monitor and conflate these movements with terrorism. Trump has sort of banned states from regulating AI altogether.
Is that going to make resistance rarer and tougher? Is it going to be sustainable?
Karen:
It’s interesting. Certainly, it’s going to make resistance tougher, but does that mean that it’s going to make it rarer? I actually think the opposite is going to happen.
You can look at the tech industry itself as a microcosm of the resistance that’s happening around the world. So tech employee organizing used to be a very popular thing. In the 2016 to 2018 period, there were a lot of employees in these big tech companies that started engaging in street protests, open letters, even unionising to hold their leadership accountable, especially during the first era of the Trump administration. That energy dispersed because of the pandemic and because of various other challenges that the tech industry was facing.
And now we are seeing a return of this resistance. Actually, one of the things that we feature on the AI Resist List is this open letter from Amazon employees, where they talk about the intersecting harm that they see their company engaging in with this accelerated aggressive AI development that is enabling surveillance, for example, and authoritarianism in the US.
It’s pretty remarkable that we are seeing this bubble up out of time like now because within the tech industry, executives have become more happy than ever before to surveil their employees, to fire them, to punish and penalize them for this kind of outspoken organizing, and yet we are seeing more of it. It’s actually in part because of the very reaction of these executives trying to snuff out this resistance, because it somehow has revealed [the true nature] and the mask has come off.
I think in the past, employees wouldn’t always protest or participate because they felt like they were actually at benevolent companies and they were working for benevolent people and they were happy to participate in that kind of story. But now it’s become increasingly obvious that Silicon Valley has become fused with Washington in these deeply troubling ways. Even though there is much greater risk to the employees, they feel compelled to speak out and to push back.
I think that’s going to be the same story for resistance everywhere, outside of the tech industry, as well. Even as there is greater surveillance, greater snuffing out of this kind of protest, it is going to actually inflame and fuel even further the desire and momentum to resist.
TG:
One of the examples is data centres. They’re set up, they drain and pollute water sources, gas turbines polluting the air. You’ve talked about this from a more perfect union outside Memphis, a poorer black community.
If Americans resist, and everything goes to the plan as you’ve laid out in these grassroots movement, do you not fear that there’ll be a race to the bottom where this kind of infrastructure is going to be put completely out of sight in mind, perhaps in the less developed world, maybe where there’s less regulation or even in places like Hong Kong, which—let’s say—is a low-resistance-low-regulation haven?
Karen:
There’s definitely been attempts by the tech industry to put these infrastructure in places where they believe there will be less resistance, and it has repeatedly failed.
I talk in my book about how Google attempted to do this in Latin America. They tried to put a data centre in a very poor working-class community outside of Santiago, Chile, and they received the biggest uprising that they had really ever faced before as a company. They were shocked that this community, which, by all accounts, has the least amount of power in the face of a multi-national rich corporation like Google. Google actually was unable to build their data centre for years and years because of the degree of push-back. So then they were like, "Well, fine. If we can’t build it in this community, we’re going to move to Uruguay." And so, then they tried and it turned out that there were community leaders in Uruguay who had read about the resistance in Chile and they then engage in the same protest and the same resistance and so Google was unable to build their data centre in Uruguay, either. These companies absolutely try to play different countries against each other, different communities against each other. They try to pit people in these ways, but it hasn’t actually been a successful strategy.
I would love to recommend a book called The Wall Dancers by Yi-ling Liu, which talks about China and the history of resistance and protest in China in relation to censorship of the government as well. It’s a very beautiful and deeply reported story that looks at how there’s actually a significant amount of agency and expression and resistance that happens underground even in places like China. That's why with the AI Resist List we had an example from China, we had an example from Hong Kong, to show that this narrative—that these places don’t have any agency and individuals don’t have any say in these kinds of issues—is not actually a correct narrative.
I've done a lot of book events in Hong Kong, and at the book events, I've found that people are just as concerned about AI development and are thinking critically about how they want this technology to shape the future and are thinking about whether or not they want to allow these companies to engage in certain kinds of reckless behaviour.
TG:
We reported last week that Hong Kong has the third dirtiest data centres in the world—after, I think, Indonesia and India. Can you tell us about the example from China with ByteDance on the AI Resist List ?
Karen:
Oh, yeah. You know what, we actually have two examples from China.
So, the ByteDance example. There was this really interesting moment, in which ByteDance tried to release a feature with their video generation model that would allow users to upload a single photo and then generate a fake video of that person in their likeness. There was such a huge backlash on social media that ByteDance actually had to suspend the feature.
There are a number of examples that we have on our AI Resist List where there was very clear actual reaction from the company, which is, the company had to actually pull back what they were trying to do. These examples are rare. A lot of the examples that we have are more focused on accountability where the cases are still in progress, like the company hasn’t responded yet and it’s just a building or accumulation of pressure. But actually in a lot of the examples that we found in China, the accountability had already happened.
The second example is of a voice actress in China who sued—I forget which company—for taking her voice data without consent to then create an AI voice. She actually won the case in the Chinese courts. Both the AI company and the company that sold her voice data to the AI company were fined and held legally liable.
TG:
I feel that there’s a much broader collective growing when it comes to AI in a recent year or two.
Aside from the companies that are backtracking and rehiring humans, or companies running out of money for their tokens or whatever, pretty much everyone I notice online is that, if some company puts out a "we now have AI XYZ" or whatever, or Coca-Cola trying to do an AI-generated ad etc., there’s just abuse in the comments.
Even myself, I am getting caught out now with some of the generative AI stuff. I was kind of blubbing quite late at night listening to this beautiful 1960s Italian song on YouTube, and then I realized it was bloody AI, and I can tell. Even as journalists, we’ve also been caught out occasionally. I guess there’s that icky weird feeling you get, the feeling of being ripped off, that only existed after you learnt what was going on. Why do people feel ripped off by that kind of content? And if it becomes totally indistinguishable from the real thing, are people really going to continue to care in years to come?
Karen:
I think often times we get a little bit caught up in this idea that if AI generated content is indistinguishable from human generated content, then that means they’re interchangeable. But that’s [not the case].
[Even if] they’re the same at a superficial level in their presentation, but they’re not actually the same in terms of what they represent. Human-generated art is a representation of a person’s lived experience and the reason why we like looking at human-generated art, listening to human-generated music and we feel that ickyness with AI-generated stuff is because we are in that moment of engaging in a piece of art, like entangling our life with another person’s life, and it feels really meaningful because we know that at the other end of that entanglement was a real person who had emotions and thoughts and experiences that were similar to ours.
So, even though AI-generated art might at some point be indistinguishable in certain respects at a presentation level, it doesn’t actually represent the same thing.
Will there be moments where people might consume AI-generated art and still have an experience that is meaningful to them personally? Yes, I’ve talked with people who feel that they’ve already had that experience. But is it the same thing as interacting with human-generated art and actually engaging with another real person’s thoughts and feelings and emotions? Not at all.
TG:
Yeah, I was quite surprised actually some of the students we spoke to at universities for a future last week were pretty cynical about using AI. This is when the journalism school converted their entire school to be AI-based. I also spoke to a teacher who would sometimes show a picture of a beautiful lake in Switzerland or something to the kids and they’d be like, “Ew, AI.” I mean, it was a real lake actually, but these kids I guess are used to quite ramshackle Hong Kong beaches.
But I wonder if that next generation, the very youngest Generation Alpha or whatever—if Generation Z were born into the internet era, there’s going to be kids coming up—would they be just utterly accepting of AI whilst the rest of us are Abe Simpson shaking our fists at the sky?
We talked about the differentiation that may be important to us (whether we can tell the difference or not), but will the next generation, the youngest, really care?
Karen:
I think that it’s exactly the opposite way around.
I think the younger generation care the most, and this was with social media as well. I was the social media generation—I’m a millennial—and it was millennials that started first asking the questions like "Did we actually enjoy growing up in the social media era? Should we be in doing digital detoxes?". All of that cultural push-back against social media, and now many people starting to buy dumb phones, happened with millennials. There was an article from the Financial Times column by John Burn-Murdoch where he looks at the data for social media adoption currently, and it’s already peaked. It’s just in decline now, and the generation that has been the fastest in getting rid of social media is the youngest generation.
So, we’re seeing the same thing with AI. Gen Alpha and Gen Z are saying things to each other in school like "That’s AI", meaning that’s fake news. One of my my colleagues was explaining to me how a conversation would go. It’s like, a guy says to his friends, “Oh yeah, I have a girlfriend in another school.” And they’re like, “That’s AI.”
TG:
Haters will say it’s AI.
Karen:
Yeah, exactly. AI is being coded in a negative way in the younger generation. Actually, the people that speak the most excitedly about AI are the boomers. They’ve already had a long and successful career. They’ve already bought their home. They’re business owners, capital owners, etc. They’re the ones that can benefit a lot from this technology. So, they’re super excited about it.
TG:
Yeah, I feel similar in that. I’ve just made it into the millennial cohort, and I’m trying not to behave like Abe Simpson. I’ve printed off our five-page guide for staff because we seek to be GenAI-free. No one sentence on our website should be generated by AI. But we also don’t want to be Luddites and we want to make sure that when it comes to processing data or transcription and experimenting, we’re not completely head in sand.
You’ve said that you’re not allergic and you use AI, but I’m super curious to know exactly what tools you use and how you might be using it. Is it mostly for research for what you do? Or you’re getting use out of it? How exactly are you interacting with it?
In fact, because of you, me and my partner don’t use it for frivolous things, because we have the environmental thing at the back of our mind. So, no cat pictures and things like that.
Karen:
So, we don’t use GenAI. I don’t use ChatGPT, Gemini or any of the tools. If a new feature comes out, I’ll occasionally test it to understand it better for my reporting purposes. But I’ve never used the tool for research purposes. I don’t recommend people use it for research purposes. The number of times that [you get errors is quite high]. Now Google Search is AI search (everything is AI search), but I think one in every 10 searches I get a really egregious factual error in the AI search. And it often surfaces that things actually aren’t even in the sources that they link to, which is very very confusing.
I don’t use those types of tools because of three reasons.
One is the ethical stance that I take. It's because I’ve been investigating these companies.
The second one is a privacy stance. As I investigate these companies, I think that these are the greatest surveillance tools that have ever been built. Every time you use it, you’re giving an extraordinary amount of data to these companies. And for me, I’m particularly sensitive about that.
The third reason is that I don’t think that any of my work actually benefits from GenAI specifically. So I can take the ethical stands without having a particular cost to me.
But I do use other types of AI tools. There’s GenAI and there’s other types of, like, transcription services. I use AI transcription all the time. But I specifically look for transcription companies where:
(1) They have a very high data privacy standard.
For my book, I use a transcription service that only did transcription on my local drive. It was not cloud-based.
(2) They are not actually using GenAI.
It's because some transcription tools have actually switched to GenAI in the back end. I just think that’s like using a rocket when you could be using a bicycle.
The other type of tools that I I use... In my book I had this particular detail that I wanted to add, where I was trying to explain to readers how OpenAI had received a really dramatic upgrade in its office after it went from a non-profit to a Microsoft-backed venture. I had noticed that their chairs had gotten significantly fancier. I took screenshots of both the chairs from their old office and their new office and ran them through Google Reverse Image Search, which is a predictive AI tool—it's a specialized AI tool and doesn’t use GenAI as far as I know. [The Image Search] gave me a match on the type of prices that these chairs typically go for online. The first office had chairs that were around $2,000 each. The second one was around $10,000 each. So I added that as a detail, a color detail, into the book to try and evoke the sense of wealth that we’re actually dealing with in these echelons.
TG:
We’ve definitely noticed this year the number of mistakes AI seems to make and those AI overviews. It’s one of the reasons we shun it basically.
I feel, particularly with Hong Kong, all this bias (if you’re kind of an expert on an issue, you will notice it more when it regurgitates these things). Particularly with the AI overviews, I feel it’s getting more deeply embedded in our tools—in our phones, in our software, etc. Sometimes quite irrelevant software, which really doesn’t need it.
So I wonder, if you’re quite careful and sensitive, churning of certain types of AI is sustainable in the... We’re going to be buying phones where it’s just baked in or it’s happening behind the scenes with what you’re using. I also fear that they may not declare these kinds of generative algorithms are going to be a play because it’s becoming such a toxic thing.
Karen:
Yeah, I absolutely think that these companies are trying to bake AI into everything because they’ve realized that just relying on consumer demand is not actually helping them turn a profit. So now, you have this thing where I pay for Google Workspace and about six months ago my monthly subscription jacked up its price not because of any feature I’m using but because Google has has decided that they can charge more simply because GenAI is everything now. I’ve disabled GenAI but I still have to pay that premium. That’s how they’re trying to get their money back on the extraordinary investments that they’re making in data centres.
I think there are a couple things that need to happen. This goes back again to the idea of resistance as a an important mechanism of accountability because without the external pressure, companies will continue to go down this path.
Data centre protest, for example, is a really critical mechanism of accountability that has started making it much harder for companies to develop and deploy their technologies. In 2025, over $150 billion of data centre projects were stalled. This, I think, is in just the US alone, according to Data Center Watch. Many of these facilities were OpenAI facilities, as an example. OpenAI recently had to shut down its video generation tool, Sora. I think there’s a direct line that you can draw from the data centre protests and that shutdown of an entire product line. It's because when you look at the reasons for why OpenAI shut down Sora, all of them were shaped by this grassroots movement.
One line was that they simply were constrained by their computing resources—okay, that’s like a very clear direct line.
A second line is that they are about to IPO and they are facing much more financial pressure and uncertainty from Wall Street because Wall Street has been observing all of this resistance and protest and is becoming increasingly worried that the AI industry can’t actually meet its promises. They’re writing about it in their memos and documents, about these companies, and they’re pricing it into their valuation. So OpenAI feels a lot of pressure to shore up and make less risky bets, which means shutting down an entire product line that they’re not really sure how to monetize.
The third line was flatlining customer demands. They just were not seeing people actually using these tools.
The constellation of these three, all shaped by collective action, was what made Sora meet its demise.
So is there going to be a a huge push from the industry to keep going in this direction? Yes. But that’s why we as individuals and communities and organizations need to be thinking about how we actively use these kinds of push-back mechanisms to shape the trajectory of AI development.
The last thing that I’ll add on this is: A lot of people think of themselves primarily as a consumer when in relation to these companies. It’s like, you either use or you don’t use these tools, and that’s the only option that you have. Again, I don’t actually think that’s the case. You can also be a resident within a community pushing back. You can be a voter. You can be a business leader that actually doesn’t replace your workers, so on and so forth. But consumer is of course a very important hat that we wear, and it is in fact one of the most tangible ways that we as individuals on a day-to-day level can help shape and vote with our feet on what we want these companies to do.
One of the challenges that we have right now with the AI industry is that the burden on the consumer is extremely high in terms of actually understanding what these companies are doing, where they’re putting their AI models, what kind of supply chains they engage in. But we’ve seen this problem before. The fashion industry had this problem, the coffee industry had this problem. As all of these supply chains mature, they end up being like consumer advocacy groups that actually audit the supply chains of these different brands, and they tell consumers "these are the values that this company is actually enacting, versus that company". That gives the consumer more ability to actually move from one product to the other. That also has to be in conjunction with the fashion industry. There was consumer advocacy, there was labor organizing, there was government regulation, international norm setting and all of this in conjunction with one another. That then created new markets for sustainable and ethical fashion brands that then created more visibility for the consumer and so on and so forth.
So we have to go on the same journey with the AI industry. That is yet another thing that we can as individuals do right now to demand that we get there eventually.
==========3700===============
TG:
As the AI models would say, you’re absolutely right.
I grew up amid the backlash against Nike 80s, 90s. I remember that, but then China came along with Shein. I wanted to talk a little bit more about China because a lot of the justification for all of the ferocity of AI development is "because China...". "If we don’t do it, they will." "They already are doing it." "They have massive government subsidies. Don’t regulate us." Is this actually justified? Is this like the space race with the US paranoid about Beijing gaining a military edge?
Karen:
I first want to go back to your Shein comment because I think this is a really really really critical point.
Right after I finished writing my book, I read Rebecca Solnit’s Hope in the Dark, which is a very short but beautiful meditation on the history of grassroots movements around the world and how to think about the role of the grassroots movement in pushing forward social and moral progress in society. One of the things that she says in the book is, people tend to have this belief: To win in grassroots movements means that you’ve reached a destination and then there’s never any backsliding ever again. Like you’ve just arrived there and your job is done. If you arrive there and then there is backsliding, somehow that means that everything that you did beforehand doesn’t even matter anymore. That's the belief. I have a lot of friends who are activists and participate in grassroots movements, and they say this is one of the struggles of younger-generation activists. It is this mentality of "if I fought really hard for this thing and then, like, Trump is elected and everything reverses back, does that mean it’s all pointless?"
In fact, no! Like what Rebecca Solnit says in her book, the point of grassroots movements is that you’re always just being the vector that pushes in the direction of the correct moral force. There will always be other forces that are trying to push back. So yes, you should absolutely expect backsliding. But the backsliding doesn’t mean that everything that you did before is pointless. It means that you have successfully got to a particular place with that action. There’s going to be backsliding and then your job tomorrow and the day after and the day after is to continue pushing in the other way. So, the fact that she has reverted back to certain types of unethical practices in the fashion industry is not in any way discounting all the incredible work that was done to shore up the fashion supply chain beforehand. There will be continued work by those same people to continue pushing and making sure that we can we continue, like advancing the fashion supply chain.
TG:
Do you not feel there is something in the whole "China threat" narrative?
Karen:
I don’t think it’s a real narrative. I mean, it it is a narrative for sure, but I don’t think that it’s based in reality.
Look at the track record that this narrative has actually had over the last 10 years. This is not the first time that Silicon Valley has deployed this narrative. They deployed it through the social media era. Over the last 10 years, what’s happened was: Silicon Valley said, “We’re going to dominate social media and we’re going to have a liberalizing effect on the world.” And the exact opposite happened. Now the most dominant social media company is ByteDance, and Silicon Valley has had an deliberalizing effect on the world.
With AI, it’s the same thing. They said, “We’re going to dominate the AI space and going to leave China in the dust.” And instead, now China has figured out how to create these open-source, much more efficient models, like Deep Seek, that US companies and US academics often prefer to use now, because it’s free. Why would they pay significantly more money for the same service instead of just downloading it from online?
So this narrative which Silicon Valley has just continued to use to accrue more power to itself has actually backfired in in many different ways. One of the things that I find, a particular pet peeve of mine, is that I think people have a tendency of thinking that anything that comes out of the US is somehow American and represents American democratic values and anything that comes out of China represents the Chinese government and authoritarian values. At the end of the day, I do NOT think any of the companies in Silicon Valley are American in any sense of the word. They do not represent the American people. They’re not acting in the interest of the American people. They’re in fact undermining the American people and in many ways undermining the American Republic. The ultimate thing that they are trying to do is just accrue more power to themselves.
With these Chinese AI models, the open-source ones that are now coming out, you can’t say that they are authoritarian because they are actually democratizing the technology in the way that Silicon Valley often says that they are doing but are in fact not. It's because you can download these tools, you can modify them however you want. They’re free. People are using them in various different ways that are completely outside of the control of the Chinese government or the Chinese companies. Anytime you use them, you’re not sending data back to the company itself because it’s open-source. So, I think that the narrative that the AI industry in the US is deploying is one that just doesn’t hold up against reality.
TG:
One more, the AI bubble. We’ve been told, for over a year now, we’re in a financial bubble looking to the dotcom boom. How much worse may the effects be if and when it bursts? Is it going to take our pensions down? We’ve been hearing this for a while, is it a conspiracy? Mark Cuban said last week that it might burst gently and AI firms might fall into categories in which they specialize. "This one is doing a bit of healthcare, that one does a bit of coding and maybe it’ll not be so bad." What do you reckon as we sign off with this obscene amount of money that’s being thrown around?
Karen:
I don’t engage in predictions really because I I think that predictions ultimately make things feel inevitable and the entire core of my work is this idea that nothing is inevitable and everything that we do today is what shapes tomorrow. We should just make sure what we do today leads to a future where we don’t have catastrophic bubble burst.
TG:
Karen, thank you so much for joining us. You’re going to be giving the keynote at the SOPA Awards this week. Empire of AI is out in paperback now. Thanks again for coming in.
Karen:
Thank you so much for having me.
Source: HKFP
