Air Date: 9-26-2026

Today we explore the layers of chaos surrounding the AI industry, including the idealistic nerds-turned-nazis, the grift, bubble, and doom-marketing business model, and what people are doing to fight back.

Full Show Notes

Welcome to this episode of the award-winning Best of the Left podcast.

Today we explore the layers of chaos surrounding the AI industry, including the idealistic nerds-turned-nazis, the grift, bubble, and doom-marketing business model, and what people are doing to fight back.

For those looking for a quick overview, the sources providing our Top Takes in about 55 minutes today include

Political Breakdown

The Tea with Myriam François

Elena OnOffline

Tech Won't Save Us

Better Offline

Blood in the Machine

Adam Conover

and Nash Keune

Then, in the additional, Deeper Dives half of the show, there'll be more in 4 sections;

Section A, The Nerd Reich - Who They Are and What They Believe

Section B, What AI Is Actually Doing To Us

Section C, The Grift, The Bubble, The Doom Marketing

And Section D, Fighting Back - Workers, Communities, and Alternatives

And now, on to the show.

in this philosophy, which you, you dub a libertarian fantasy in which civilizations collapse as a feature, not a bug, h- what happens to the rest of us? 

It's not good. I think we're either servants or serfs, or worse, dead. I would say look at what's happening to the poorest people in the world who are dying because of the USAID cuts, right?

That will be the majority of people on the planet in 20 years. Carried out by Elon Musk when he was doing those, yeah. Carried out by Elon Musk, which didn't even save money, by the way. But what we hear a lot from the AI industry is that AI will kill all of our jobs, but that will, it will also create this super abundance, and we'll all somehow get a share of the universal extreme wealth, to use Sam Altman's phrase, that will be created.

The problem with that is that we already have abundance, and it's not being fairly or rationally or logically distributed. Yeah. So there's no reason to believe that that will be the case if AI can even do the things they say it's gonna do. I'm not saying that I buy into that idea that AI has to replace everything.

But the same billionaires who are promising abundance all tend to be building bunkers around the world and buying islands- Yeah ... and creating new hideouts. They think the future's gonna be very dark. And I think what Peter Thiel's been able to do is create a cult of apocalypse capitalism that sees the end of civilization coming and hopes to find some way to profit from it and thrive in some fantastical science fiction future where the rest of us aren't really on the map.

So you you describe, Well, you talked about how PayPal is part of this, hedging into this- Mm-hmm ... digital currency. obviously that's gone way further with crypto. But I, I wonder, you, you also describe his investment in now Vice President JD Vance as similar to how he invested in the venture capital world.

Talk about that- Mm-hmm ... 'cause he's been a patron of Vance, right? 

I would say he's the creator of JD Vance. 

Mm. 

My chapter on JD Vance is called The Unicorn, and in venture capital terms, a unicorn is a billion-dollar company. You make an investment in something and it becomes worth a billion dollars eventually.

At every step of JD Vance's adult career, it's been Peter Thiel guiding the way, promoting him, and funding him. JD Vance left law school and decided not to be a lawyer, because of Peter Thiel, went to San Francisco and became a venture capitalist at a venture capital firm co-founded by Thiel. Went back to Ohio, started his own venture capital firm with funding from Thiel, even though Vance was a mediocre venture capitalist.

Ran for Senate. Peter Thiel spent $15 million, setting a record for individual spending in a Senate race up to that time. And then he helped get Vance on the ticket with Donald Trump, even though Vance had previously called Trump a cultural opioid, an opioid of the masses, and America's Hitler. So it's really, I think a, a, a Senate candidate in Michigan recently called JD Vance Peter Thiel's boy, and that's a very apt description, and now he's a heartbeat away from the presidency.

And do you think he shares Thiel's belief that, basically freedom and democracy are incompatible? 

Without a doubt. In fact, during his 2021 Senate race, he was quoting Curtis Yarvin. JD Vance was quoting Curtis Yarvin- Oh, another 

character I would love you to introduce us to . 

Yeah. 

Yeah. 

Th- this, fascist political guru who is also funded and supported and promoted by Peter Thiel, and specifically Vance said during this 2021 interview that he believed Donald Trump would return to the White House in 2024, and that when he did, he should purge the federal government, and he should defy the law and challenge the courts to force him to do anything.

Mm. And he was right. Donald Trump did return to office and do all those things, and he did it with JD Vance by his side. 

Yeah. Talk a little bit more about Curtis Yarvin, who is this guy, and, what is his role in this broader ideology? 

Curtis Yarvin is this mediocre computer programmer from San Francisco who dropped a tremendous amount of acid in the 1990s and the early 2000s, and then wrote a fantasy version of fascism for the 21st century while living off of his wife and his mother.

And he had this pseudonymous blog where he wrote as Mencius Moldbug, and this came to the attention of people in Silicon Valley, and Peter Thiel became a, a big fan and supporter. Mm. And the basic thesis of Yarvin is that democracy is dumb. American democracy r- should be replaced with a dictatorship led by a CEO, and the country should be broken up into these little corporate fiefdoms he called patchworks that would be run by authoritarian corporations that even had the power to execute their citizens for no reason, and where everyone would be under constant and total surveillance at all times.

 

To understand this thing that we're calling AI, first of all, you have to understand that we call something different AI every five years, and we've been doing that since the '50s. The thing we're calling AI now is something built on something called theory-free inference, which is when you try to understand how A might cause B without understanding why, right?

So you're just going out there in the world and you're going "Oh, I saw A happen and then I saw B happen. A is probably causing B. I don't know why, but that's okay. I knew that-- I know that if I do A, B is gonna happen." Before we had theory-free inference, the AI that we built was sometimes called symbolic AI, and it was built on these things called world models, which are exactly what they sound 

You want AI to, oh, I don't know, simulate hair on a, in a animation. So you explain to the AI the physics of hair. You, you write a program, and then you give it a bunch of training data, and you give it a bunch of hair, and then you ask it to make new hair, and you ask it to change the way the hair is moving in a shot, and you, you can do a lot of things.

Now that's hard, right? Explaining the physics of hair is a very complicated, gnarly problem, and so is explaining the physics of cancer and explaining the physics of buildings falling down in an earthquake. And so these researchers, 10, 12 years ago, they had this interesting idea. They said we can observe a lot more correlations in the world than we can explain."

And of course, correlation is not causation, but causes and effects are correlated What if we just ask the machine to look at all the things that happen and look at all the things that happen before those things happen and see if they can find things that are correlated? We can observe so many more things that are correlated than we can explain.

If we just know A causes B, and we just need B to happen, and we just go out and do A, maybe it doesn't matter if we know how it works. So if w-we know that when this word occurs, then that word should occur, maybe we can make plausible sentences. And this pixel should follow that pixel, maybe we can make plausible images.

And not only did it work, it worked really, really, really well. although, of course, it's reaching limits too. and there are always intrinsic limits to this theory-free way of running, an AI, of running an automation system. the big one is it just doesn't understand. It doesn't have any understanding.

It is just using statistical models. It's guessing words and pixels and equations and whatnot. Again, that gets us further than we thought it could, but it's still got limits. To explain those limits, I was on a podcast with a guy who was a big AI believer who, shortly thereafter, took a job at a big, AI company, a revolving door.

And he said the most AI-pilled bro-y thing imaginable. He said, "I, can predict what my wife is gonna say sometimes, and the, autocomplete on her phone, the AI-powered autocomplete on her phone can also predict what she's gonna say. Doesn't that mean that the AI understands her as well as I do?" And of course, you feel for this man's poor, poor wife.

But to give him his due, it is true. I can sometimes predict what my wife is gonna say, and she can predict what I'm gonna say. Partly that's because, everyone, we repeat ourselves. But also, it's because we understand each other, and we know where one another are coming from. I cannot only com-- predict what my wife is gonna say, I can predict what she wants for dinner on a-- after a hard day, not just because I have a statistical lookup table.

It's because I know her. now the AI can also predict what I'm gonna say and predict what my wife is gonna say. There-- I've got a autocomplete on my phone, and I start typing a message, and it makes reasonably good guesses. Now, that works well, but it doesn't fail well. It fails very badly. If my wife were to say something to me that she'd never said before, "Cory, I want a divorce," I could maybe predict why she said that.

I could understand it. I could maybe make some guesses about what I could say next that might avert the divorce. whereas the chatbot, all it can do is look at all the times someone said to someone else, "I'd a divorce," and find the average thing that people say next That's clearly not the same thing.

And so we can get pretty far with theory of inference, but only so far. So this approach, right, where we never have to understand the world, where we just observe things and then observe things that happen after the things we observe, and then find the correlations and never understand why A causes B.

It's just enough that A sometimes causes B, and whenever we want B to happen, we go out and we do A. That has its limits. And, I'm a materialist. I don't believe in souls. I don't believe in spirits. I think everything we call consciousness happens in our bodies. I will stipulate, 'cause I don't understand quantum mechanics very well, that sounds maybe some of that stuff happens outside of our bodies using something to do with quantum mechanics.

But all that is to say that consciousness is a thing that happens with atoms and subatomic particles. And so someday we might collect some atoms in a way that is conscious, but we won't do it by guessing more and more words. thinking that if you guess enough words, if you put enough words in the word guessing program, it will become conscious, is thinking that if you breed horses to run fast enough, one of the mares will give birth to a locomotive, right?

It-- This-- There just isn't a path from A to Z. Now, maybe we'll write software that'll help us figure this stuff out using coding assistance, but that's just a plugin for your programming environment. We've had new plugins for programming environments for as long as we've had computer programming.

They're great. No one ever said this program, this, plugin is so good, we should put all the programmers in a wood chipper. It's just fine. the, the-- There wasn't a drug that, a cancer-curing drug that was created by an LLM. There was a cancer-curing drug that was created by a cancer researcher who used an LLM.

There has never been a drug created by a cancer researcher that didn't have a computer or some other external machine that helped them do it, right? It's saying that it, the cancer-curing drug was created by the LLM is saying it was c-created by a microscope or a centrifuge. It was the human who created it 

So I get why the tech bros who are running these companies would wanna sell us the, fear, mongering version of the story that there's a really intelligent super AI coming for all of us, not just our jobs, but potentially for, the, the, the greater existential question of they'll take o- take over and enslave us all.

but the bit I don't get then is why the whistleblowers are also saying that, right? Yeah. I was reading that someone, in the Washington Post, there was an article about someone at OpenAI, r- advanced researcher saying, they're quitting- Yeah, Cassella ... because of this. So it's I...

what is their incentive to also repeat this narrative that actually the AI's getting way too powerful, way too smart, and way too out of control? 

Yeah, I think they have AI psychosis. I think they're they've spent so long going to the bathroom, shutting the door, holding a flashlight under their chin, and looking in the mirror and going, "AI," and scaring themselves silly that it's just it's stuck.

You think they're believing their own hype? 

I think that they are high on their own supply, and it's n- It... The fact that you are smart about a domain does not preclude you being incredibly wrong about very adjacent domains. My, my favorite example of this is Watson of Watson and Crick, who won the Nobel Prize for, identifying the spiral f- shape of the DNA molecule.

he's, at the time, one of the most important geneticists in the world. We should note that he stole his research from a female colleague kn- called Rosalind Franklin, who died and never got the Nobel Prize or recognition. But Watson devoted his life to eugenics, to going around and saying, "Genetics proves that people of color are stupider than white people, blah, blah, blah, blah, blah."

And, he made a bunch of claims about what could be discerned from DNA to, to back this up. Now, we have a new, relatively new discipline called computational genomics, which is where we sequence the DNA of a lot of people, and then we find out things about population scale, genetic data that tells us what, if anything, the idea of a race correlates to, the extent to which things professional achievement are correlated with things in your genes and so on.

And it turns out it's just nonsense. It's just empirically nonsense. It's just it's just a thing that is absolutely not true. And there's a great British scientist and science communicator called Adam Rutherford, wrote a great book about this called How to Argue With a Racist. He's a computational genomist, and before Watson died, he went around the country debating with Watson and just making him look an absolute idiot.

So this guy who won the Nobel Prize for discovering the DNA molecule was the wrongest person in Britain about DNA at the same time And so sure, you can have insights into how to make a chatbot that does a bunch of compelling things, but it does not safeguard you from becoming just a absolutely deluded maniac about what the p- possibilities of that chatbot are.

OpenAI announced its AI solved one of the seven hardest math problems on earth, a Sphinx riddle of a problem. It's a millennium prize problem, unsolved for ninety years, a million-dollar bounty since two thousand, which I think when a million dollars meant a little bit more, but this is not a story about inflation.

So the two mathematicians that OpenAI supposedly beat say that OpenAI only pulled it off after hearing about their unpublished work. When one of them said that he would go public, an OpenAI scientist asked him, quote, "Why would you ruin your career?" Menacing tone implied. Here's what happened. Two mathematicians, Tristan Buckmaster of NYU and Levent Alpoge, who works at Anthropic, spent about a month on this Navier-Stokes problem.

It is an equation behind the weather, aircraft, blood flow, which sounds an Earth, Wind & Fire cover band started by some grad students. Now, the ninety-year-old question in plain English is, can a smooth flowing fluid suddenly break down to a point where the math says that the speed of that fluid goes infinite?

It doesn't really matter to the story, but it is interesting. This week they published proofs formally verified, so a computer checked it line by line. They didn't crack Navier-Stokes itself, to be clear. They solved a major stepping stone. And Terence Tao, who is a Fields medalist and your favorite mathematician's favorite mathematician, he basically called it a remarkable achievement and said that he looked forward to them going further.

But instead of them going much further, they got lapped by big tech. So the suspicious part of all of this is that mathematicians were working on an obscure route that almost nobody in the field was using. OpenAI, by its own admission, only started its work after hearing a rumor that the humans were close, and they showed up on that exact path, brambles and all, the same week.

OpenAI threw ten thousand AI agents at the problem for eighty-eight hours straight and now claims that it went further and cracked the full thing. So a long weekend's worth of labor on a almost century-old math problem. OpenAI's proof is machine checked, but no outside mathematician has vetted it.

Quickly, before we go further, why does a chatbot company even care about a ninety-year-old fluid math problem? Math is the one place that AI cannot fake it. The proof checks out or it doesn't. It's black and white. It's the cleanest way to prove that your AI actually reasons, and it is a teaser for the real prize.

An AI that does original math is a step towards one that does original science. Drug discovery materials, the trillion-dollar stuff, curing cancer. So the ugly part happens in the receipts, the DMs, where it always does. OpenAI offered Buckmaster a deal, write up the win yourself, credit our model as the thing that solved it, and we'll publicly call you guys the closest humans to the problem.

The proximity-based consolation ribbon by letting them just stage manage the whole performance. There was one condition, though. They had to drop Alpoge from the research because he works at Anthropic, and they didn't want a rival's name near their big moment. So when Buckmaster said no, and said he would go public, a true friend, that is when the OpenAI scientist, Sébastien Bubeck, told him, "Why would you ruin your career?"

Bubeck has since apologized for that line, and he disputes Buckmaster's account, but he did say it. Here's the thing. Nobody has proven that there was any theft here. Buckmaster himself says, "I have not seen the proof." OpenAI is defending itself very hard. They're saying that it acted with integrity, offered the humans the prize.

OpenAI was the one that was actually threatened. It's a he said, she said, and there's no smoking gun. But the thing that OpenAI is selling, the machine solved what humans could not, is always supposed to be the story behind these big, monumental achievement breakthroughs. And that basically falls apart because the humans it beat were using AI, too.

But the headline, which is AI solved what humans couldn't, is a little bit too simplistic. Buckmaster and Alpoge were obviously using AI as well, Claude and OpenAI's own tools. The difference is that OpenAI could spend, roughly ten thousand agents, throw them at the problem for nearly four days, and spend millions of dollars doing it.

And that changes the whole contest. A small team can have a great idea, and a giant AI lab can take a promising direction and just hammer at it on a scale that nobody else can afford. So can AI beat mathematicians at this point? Yeah. It's just whether original research is becoming a resource race.

Who has the most money? Who has the most compute? And if one side has two mathematicians and a laptop's worth of AI, and the other can spin up ten thousand artificial researchers overnight, then it's not about who had the better idea. It's not even who got there first. It's who could sandblast it with artificial intelligence.

​

Musk is, is not a libertarian, that if you look over the course of his career, he's consistently instrumentalized the state as a source of power and profit in various ways. as a, as a backstop, as a funder of basic research, as a provider of loans and subsidies, and really above all, as a customer and a client.

You mentioned SpaceX, and, SpaceX gets its start as a key government contractor in the early years of the War on Terror, and now has, become a, a very important partner for the US military and US intelligence agencies in getting s- its satellites into low-earth orbit. I think, one way to read that is through Musk personally, and another would be to say, well, what are the shifting paradigms, both technological and financial, that drive the evolution of the Valley from the '90s onwards?

And if you think about high point of so-called cyber libertarianism emerges as a, as a better known formation in the, the 1990s into the 2000s, is associated with figures Peter Thiel And if you think about how people are making money on the internet in this era, both the Web 1.0 and the Web 2.0 era, it's websites and apps.

It's consumer-facing. Obviously, they develop a surveillance capitalism model where it's just all about harvesting user data. But the relationship to the state in this period is a bit attenuated. of course, there's always a relationship. DARPA is always important. you have the NSA doing collaborations with, with the big tech firms around mass surveillance.

But certainly when put alongside earlier eras of the Valley's history, if you think about its origin as a semiconductor manufacturing zone for the Pentagon, the relationship with the state by the '90s and 2000s has become, a bit more distant. Everything changes with the generative AI boom after 2022.

Now you really need the state as a close partner, not just as a client. think about the Pentagon's endless appetite for AI kill chain tools, as evidenced by their partnerships with Anthropic and Palantir. But also crucially, probably even more importantly, as a partner who is going to help accelerate the data center construction boom.

And you've seen this quite dramatically in the case of the current Trump administration, which has rolled out all sorts of policy moves designed to speed up construction, whether it's about providing federal public land for data centers, rolling back environmental review, trying to secure energy infrastructure, and so on.

And that's, if you are, Sundar or Satya or these guys who, at the end of the day, don't strike one as particularly reactionary ideologues but are quite rational capitalists, you actually now have a very strong material incentive to partner with the state. It's not just about trying to avoid, Trump's bad temper and the possibility of retaliation, which every capitalist now has to worry about, but you really need a partner for this big, big investment push.

Here, as elsewhere, Musk gets there first. for Musk to be developing this type of relationship with the state as a client with SpaceX as early as 2002 shows us, I think, once again, that Musk is not only a indicator species for broader developments, but he's often someone who arrives at a new political economic inflection point a bit earlier than his peers.

So by the time everyone else is partnering with the state, turning to hard tech, turning to infrastructure, he's been doing that for decades. 

Just to, demythologize him a little bit, also the case that he's also often coming late to sectors, right? and, and is still lags, for example, and as Paris, from the book you wrote about this, but his attempts at doing fully self-driving automobiles have been, promised serially again and again, and have never actually panned out, even compared to his competitors.

And xAI remains a serious laggard In comparison to OpenAI and Anthropic. So part of what I think is important to home in on is what Ben is pointing out, that it's really this, this willingness to find a partner in the state that made him special, and he's always managed to eventually get there in most cases.

And now arguably, we're at the doorstep of his biggest bet yet with SpaceX, which a hundred percent relies on state as client, really primary client, but then also with the hope of creating whole new market sectors around, satellite to internet, satellite to cellular, satellite to device, and supposedly orbital AI.

So it's always this combination of overpromising, but then backstop by Uncle Sam, as he once famously put it. 

Yeah. you even see that with the AI push, right? the, the contracts with Grok and, how that... I don't know. Even if you really wanna bring a generative AI tool into government, is, is Grok really the one that you would be contracting with?

But- Probably not, not unless he makes it a condition of the, the deal, which is what he's been doing, forcing banks that want to be part of the SpaceX IPO to download Gro-Grok just to be able to be part of it. So that's one thing that we thought about, especially in the early part of the book, is this interesting tension in Silicon Valley operating practice, which is both sure that there's always a disruptor hard on your heels ready to overthrow you, but then for good reason, completely fixated on how to lock in your own incumbency and make sure that you are not the one who is displaced.

And you could see the whole-- this is really what Nick Srnicek's work has been about recently, is these strategies of entrenching market dominance and market share, which often mean, turning yourself into a conglomerate, acquiring competitors, somehow getting-- squeezing out competitors. But ironically, working against the supposed core ethos of Silicon Valley itself by, stopping disruptive innovation and making sure it doesn't happen, thus, arguably making the quality of the things you're delivering worse over time as well.

I've now had quite a lot of emails about the post by Evan Hubinger and Jacob Coxon of Anthropic saying that they, and I quote, "earnestly believe AI could kill all humans with a higher than ten percent chance that it happens within the next decade." I wanna lead by saying that these men, regardless of their intentions or cynicism, are beyond loathsome.

They are disgusting to me. I find them vile, and I find it equally vile how many people are falling for their garbage. If they truly believe what they're saying, the idea that they're continuing to work on a product that could destroy humanity makes them wannabe war criminals. Coxon, who recently quit Anthropic, did so in an extremely public way that has been covered by much of the mainstream media, claims that he quit, to quote The Wall Street Journal, because he "doesn't want to participate in an industry-wide rush to build AI systems that can improve themselves," which as I will refer to later, is referring to recursive self-improvement, which nobody has proven actually is possible.

In an interview with Wired, Coxon spoke at length without really explaining what it was he was scared about, outside of one moment where Max Zeph asked for specifics about his concerns, at which point Coxon incorrectly describes the hack as "AI doing this all of its own volition," referring to the Hugging Face attack, only for him to respond when pressed about whether this is companies just moving recklessly fast, that he didn't want to focus too much on the Hugging Face attack because there's plenty of evidence that the companies don't know how to align the models properly.

In other words, the moment that Coxon was asked to get specific about what the companies are doing wrong and where they're making mistakes, he punts to a non-answer. By the way, Coxon's solution to all of these problems is that the AI labs should agree to not rush towards recursive self-improvement.

That's still theoretical term for an AI that can train itself that everybody's talking about it's real. In other words, Coxon is suggesting that Sam Altman and Dario Amodei put their clammy hands together and agreed to delay breeding the Grinch. We're not getting a Lorax, folks. We're not making Shrek.

We will not let the Fairy Godmother into your house, okay? Godzilla will not happen. Yet the most laughable part of the interview was when asked about why he kept working at OpenAI and Anthropic for years, when Coxon said that it's also not hyperbole that we could cure cancer, because OpenAI, by stealing someone else's work, I'm not getting into it, it's all over the subreddit, solved the Navier-Stokes problem, adding that there really is no reason that we, referring to the AI industry, can't transfer that to scientific domains biology.

The thing you say when you're just making shit up for attention, and you're wrong. You're just fucking wrong. The... Those are two very different things. A mathematical problem is not the same as biology. What are you talking about, Jacob? Why are the journalists just printing everything he says without pushing back?

It drives me insane Nevertheless, people have been saying, "Oh, oh, well, he's leaving 'cause he's so concerned. He's gotta tell everyone 'cause he's so concerned. He's so worried." No, no. The piece ends with Coxon saying that he'd to do some independent commentary on where things are going, which means a Substack, and that he'd to do something AI 2027 or move into some accountability role.

Yay! Yay, there's the grift. Whee! Congrats everyone. Congrats on helping Jacob Coxon get a new job. Congrats. Congrats everyone. You did it. You helped elevate a real piece of shit Okay. My dear friends in the media, my esteemed colleagues who allegedly have been doing this for years, Anderson Cooper, CNN, Wall Street Journal, all of you, everyone, how many fucking times are you gonna fall for this?

Why do you believe these people other than that some of them are saying vague and scary stuff and that they happen to be at the companies? Why do they never really say what it is they're scared of? And why when they even try and get specific, not that they do very much, do they never blame the companies?

I say it again, nobody has actually achieved recursive self-improvement, nor have they shown any proof that it's even possible. But the fact that Coxon is bringing it up in the run-up to Anthropic's IPO and as the rest of the AI hogs oink about it nonstop makes it impossible for me to believe that Coxon has anything other than the most cynical intentions.

People say, "Oh, he gave up his options in Anthropic," as if that matters in any way, shape, or form. He was barely there a few months. He probably didn't have that many and saw an opportunity to get a bunch of attention. He was at OpenAI for years and likely has a ton of options from there too. Why in the world does this keep happening?

This is how bubbles form. This is how grifters grift. It starts and it finishes with the media industry giving up on their jobs. It starts and finishes with the most important journalists in the world failing, and I think everyone failed here, and I am disgusted and outraged to see this happen again.

It's exactly what Matt Schumer did a few months ago with Something Big Is Happening. It's the Citrini memo about the global intelligence crisis. It's the same shit. It's science fiction dressed up as fact, and the fact it works is only because, to quote Ed Elson, we have this worship of the wealthy where we believe that the people at the companies and the people will move around from these companies in a way that's only honest, in a way that's only true, when in fact, these are some of the most cynical people in the world, which is pretty obvious in the fact that they go, "Oh, I'm afraid that this stuff is gonna destroy the world," but they keep working on it.

And I don't give a shit if he's not working at one of the companies. If he goes into an alignment role or a nonprofit for this shit, it's the same thing. It's a way to keep milking the cow. And Coxon's entire argument centers around this nebulous idea that things are moving really fast, but never really crystallizes on what it is that's moving fast, nor does it ever hold the companies in question accountable.

the impetus that we got for the Luddite Lab was effectively that we weren't really seeing any resource within either the labor movement or the tech justice space where there was a a playbook or a set of strategies that workers could use to push back against AI at work.

we've seen vague AI principles from o- other labor organizations. in the worst of it, there's been, actual partnerships between the AFL- with, between, the AFT, the American Federation of Teachers, and the companies themselves. Although they're Working that back a little bit, but not really.

but then there have been really big, huge wins against AI, just saying "No, we're not gonna have this," or, "We really want more, governance of the technology." So what we formed, we formed this project at DAIR, the Distributed AI Research Institute, where I work, and we formed this project to develop a resource hub, which featured three parts.

One of them is a case... a set of case studies that we're still developing new ones. Those case studies would be lessons from existing labor organizations, on how they've been able to fight and win against AI. And so we have four case studies from three unions, in journalism, healthcare, in the tech industry.

We also have research primers. One of them has to do with the AI productivity myth. It's the first one we published, as well as a set of five myths around that. And then a resource library of, really great stuff that partner organizations have produced, both analysis and also, also research. And then we also have an office hours process where, if you're fighting this at work with other folks, whether you're in a union or some other type of worker collective, you can, you can set aside some time with us, and we'll chat with you and, and see, if there's some ways that we can help you out and, and work together.

And so what we found, talking to right now, we've talked to... at this point, I, I, I've been meaning to go back and kinda revisit everything that we've done in the past two years. we've talked to hundreds of workers, we've talked to dozens of unions, and there's three top line things that we can highlight.

So the first is that, when folks are fighting this, it's often, not about AI. it's about AI, but it's not about AI. And I think what that means is that- I think a lot of the kinds of harms or the kinds of things that are suggested by both people on the right and the left is that the threat is mass worker displacement.

we're just gonna automate everything, and be- and, and we need to be very protectionist about our jobs. And the more extreme version of this is, the existential risk one. Well, we're not only gonna just automate all jobs, we're just gonna kill everybody. and this is stuff that, Bernie, for instance, has bought into.

Actually, he and ... I just saw he and, Senator Cesar, just proposed a, a, a bill that was regulating autom- artificial super intelligence. so this is a thing that is nonsense, but is, taken the popular imagination both of legislators and of, a lot of other people you would consider to be more discerning.

So- 

And we should say it's useful nonsense to the AI 

companies- 

Exactly, yeah ... 

the point 

you make, yeah. Yeah. Because it imbues the whole project with this, this power- Exactly ... and inevitability, yeah. 

Exactly. So one of the findings is that it's about AI, but it's not about AI. Effectively, it's, there is lots of different threats in the workspa- workplace from technology, from automation, and those are continuous from a lot of prior h- fights, thinking all the way back to, the factory line and, and, and, and the development of, of, of Fordism.

And so thinking about that, it is highlighting not talking about AI threat, but thinking about technological threat and what it means to control that and have some, some ways in which workers have a say. So that's number one. The second thing that we're learning is that, and it ties into the first, it's about control and oversight.

and so workers, are happy to explore different technologies in the workplace, if they find them to be actually useful and to be something that supports their work processes. but they don't wanna do it in a way where it's going to be mandated, where it's going to lead to a place where it's going to make their work more precarious, or it's going to lead to less dignified working conditions.

And so one thing to really think about is it, it's not just about work, it's about dignified work. what does it mean? And dignity is a complicated concept if you break it down, but more simply, it's about feeling you have control and that you're being respected on the job, 

and that includes being compensated well, having set work hours, not, not having to talk to a goddamn clanker, it's- These are types of things that people, are basic types of, baseline things. So people, not necessarily against technology, but they are against these kinds of impositions.

And, this is why we adopted the, the, the name Luddite, as, as you very well know- ... and I'm sure your listeners very well know. Luddites were not anti-technology, but against these types of overbearing, types of technologies that ruined social relations and disrupted, communities- Mm

of, of trade and commerce. 

Yeah. 

And then- 

And- 

Yeah ... 

the, yeah, just as, yeah, the AI is just such a uniquely powerful tool for undermining people's dignity- Yeah ... on every front, right? Yes. just stemming from its, found- foundational its operative use in a workplace where it's just "Oh, this, this, this, th- this chatbot can, can do your job for you."

Yes. that, that assumption by your manager is inherently d- demeaning and st- and it strips workers' dignity. So I think that that, I, I, and I, I'm glad, I'm so glad you brought that up, too, because I think it's just such an under-discussed element of all this. yes, there are real threats to the material conditions and ability for a lot of workers, especially creative workers who rely on producing text and images and that thing for their, for their living because, now we have this automated generator.

But undergirding this entire project is that. Is that, this, this, this automation, this text generator, this, this extruder can, can replace you, in any way, and that, it's just such a, it's, it's such an insulting premise for a lot of people, and I think that's what we're seeing a lot manifest in, in, in these workplaces, in the Glassdoor data, in the the interviews that you and I have done.

So yeah, I'm glad that you, you, you all have, have really underlined that at, at the lab over there. 

Yeah, absolutely. yeah, no one wants to hear that they are going to be replaced, and then to add insult to that particular injury is It's not actually able to replace what the substance of the work is.

Yeah. it is, it, it... I made a, I made this joke that, and maybe some of your listeners will, will recognize it, but, I ma- I, I made it in a, in a prior draft of the book where there used to be this shirt from the company Think Geek which s- which was, "I will replace you with a very s- small shell script."

And it is a particularly nerdy thing where it suggests your job is only, it could be automated with very little effort. And that's not the substance of, many jobs. many jobs re- require relationships, they require skill, nuance, balancing all these different types of things, and it's not that.

And, and, and, and just to suggest that it can be replaced with synthetic media is, is a real insult to those workers. And I, the, and the last thing I wanna mention that, that we're also learning just in terms, and this is specific to union workers, which is even if union workers are able to get some bargaining language into their contract, it certainly does not, end there.

AI isn't just starving independent publishers of traffic, it's literally destroying the sites themselves.

The writer Stephen Follows recently covered what happened at the site the-numbers.com, and this story is terrifying. The Numbers is a site that has tracked and reported on film industry box office receipts for 30 years. Even if you've never used it, you've definitely been impacted by their work. The site gets 8 million visitors a year.

It's relied on by journalists, the film industry itself, and even the Guinness Book of World Records uses its data. It's considered the definitive authority on box office numbers. But, recently, a couple new sites have also started leaning heavily on The Numbers' data: Kalshi and Polymarket. They use the data to settle bets, and that is the problem because the degenerate gamblers who use prediction market sites have started sending swarm after swarm of AI scrapers at The Numbers to try and get the box office receipts early so they can insider trade off of them.

Hey, remember when Martha Stewart went to prison for that? Well, justice for Martha, this is way worse. These AI tools have hammered the-numbers.com so mercilessly that its traffic became 90% AI bots. That was so much traffic, in fact, that its servers were overloaded and the site had to literally shut down.

And when the admins were finally able to relaunch, they had to do so with a bare bones version of the site. They literally had to remove hundreds of thousands of pages of historical film data because they could not protect that much legacy web code from being hacked by AI bots trying to get in. And that means that the-numbers.com, which is still trucking, please go support them, they're good people, well, their site, though, is not as useful for humans as it used to be because AI ruined it.

And The Numbers isn't the only site that this has happened to. Multiple invaluable small websites have been hit by swarms of AI bots and have had trouble staying afloat. the vast network of small independent websites is part of what made the internet great. It's part of what got us around corporate control of the media.

It's how generations of people have expressed themselves and shared information they cared about. And it is a genuine tragedy to see that network crumble under the weight of AI. But of course, the independent web, it was already being swallowed up by monopolist websites social media and Reddit before AI, right?

Well, what's crazy? Now even those sites are getting by AI. okay, let's talk about Reddit for a second. Remember, the greatest thing about the internet was that it gave real people a place to get together and talk about our weird shared interests. And most of that activity used to take place in millions of forums, message boards spread across the web.

But as a little site called Reddit began to grow, it started centralizing all those communities in one place. Just Instagram and Facebook centralized social media and YouTube took over online video. And today, if you wanna talk about your hobby or see some weird-ass memes or find community with other people who have the same niche problem you do, you're probably gonna go to Reddit first, right?

In fact, Reddit became so famous as the last place where real people could be found online, that it's become common to add Reddit to the end of our web searches whenever you need to find an answer from a real person. And it's not great that one company has monopolized the space that thoroughly, but hey, at least Reddit's still full of real people, right?

You can go there to look for a date with this mustachioed cutie who apparently loves goats, or brag about your new sewing project, or even to find out if that thing on your butt is cancer. I thought it'd be nice to not show you a picture of that one. That subreddit is a little tough to look at. But honestly, that's what I always loved about the internet, right?

That's the best thing about it. It's where real people can be their real weird selves and talk about their real weird lives. And that's what Reddit was until AI rolled in. Because remember, LLMs need massive amounts of training data, right? Text, images, ideas, grammar, sentence structure, and they need it from real people.

AI is a parasite that feeds off of our authentic humanity in order to attempt to reproduce it. So as the biggest remaining repository of raw humanity on the internet, Reddit is invaluable to AI companies. Reddit is now the number one source cited by AI chatbots. That means that your hobbies, your advice to your fellow human beings, even your weird fetishes and moles, they are all being ingested by Gemini and barfed back up to you as an AI overview.

And Reddit knows this, because guess what? Google is paying them 60 mil a year to make it happen.

Now, it's bad enough that Reddit is selling all of its users' activity and information to AI companies, but they are also destroying themselves in order to do it. Because remember, anyone can post on Reddit. And since all those companies out there that wanna sell you crap know that AI is trained on Reddit, and they know that AI is quickly supplanting real web search, well, they have started secretly planting comments and posts on Reddit that seem real but are actually just promotion designed to trick the AI into saying what they want it to say.

Oh, and it's not the nice, friendly companies doing this. It's the peptide and hormone companies. One subreddit even had to ban posts about peptides entirely because this practice was so rampant. In fact, there are now entire AI businesses dedicated to doing nothing but spamming Reddit, RedRover, which advertises, quote, "AI agents that monitor subreddits 24/7 to join relevant conversations and suggest your product to users in a natural way."

That means that instead of a place where real people get together, Reddit is quickly turning into a nest of AI chatbots talking to each other in thinly veiled advertisements. "Explain to me I'm five why Hulko supplements work so well." "Take these up-votes, good sir. Hulko supplements also taste great."

Why are the bots typing? I don't know. Why are they, why are they typing? Okay. what this means? It means that because of AI, the dead internet theory is true. If you haven't heard of it, dead internet theory was originally a conspiracy theory. It was first posted in 2021 on a tiny, super retro-looking chat board.

Ah, look at that screenshot. The old days. Brings a tear to your eye, doesn't it? By someone called IlluminatiPirate. And this user's theory was basically that most of what we think of as humans on the internet is actually just bot-generated bull and paid media influencers, all done in order to, quote, "manufacture consumers for an increasing range of newly normalized cultural products."

Now, a few years ago, maybe that sounded a little far-fetched, but today it's real. According to Cloudflare, 60% of all internet traffic today is bots. 60%. The internet used to be a place where you could find a real human being to talk to, or you could read a real person's freaky fanfic, or their genuine thoughts about model trains, or their actual experiences in their body, or their actual experiences out of their body if you were reading a trip report on Erowid.org.

But now, the internet's turned into a place where we passively receive marketing, propaganda, and spam generated for us by AIs run by massive corporations, just the bad old days. In fact, it's worse. Because today it's not just on newsstands and TVs, it's in our pockets, following us around everywhere, posting on our sites and wearing the skin of a human, pretending to be something that it isn't.

And this is why I'm an AI hater. Not just because its writing is hack and it's hastening climate change, and the flyers for local businesses it generates are way too busy, okay? It's because it is destroying the human internet I love. AI was created by a few rich billionaires for their benefit, and they are using it to muscle into places we built for ourselves for human connection, to suck up all the poetry and art and conversations and arguments, and 100,000-word reviews of direct-to-video My Little Pony movies that we made, to chew them up, spit them out, and sell them back to us for the cost of a Gemini subscription.

And they're destroying our internet to do it. 

If AI were ours, communally, then the spread of AI would free us. AI wouldn't be about just reconfiguring your 40-hour workweek. you used to spend 40 hours coding, now you spend 40 hours prompting Claude. No. Now you would just spend 15, 20 hours prompting Claude, and those extra 20 hours being a human, unbound by work responsibilities, no longer shunting yourself into the constricted form of yourself that you have to shrink into for work.

In that world, as AI got better, so would humans. In the world of mostly automated luxury earthbound communism, humans would achieve completion. Sometimes when you talk about eliminating work, people will say things "Ah, but work gives life purpose." Fucockta stuff that. At best, this is cope. People who work 40 hours a week for 40 years of their lives, and w- it would be nice if that wasn't meaningless.

At worst, this is Republicans justifying putting work requirements on elementary school lunch programs. Work doesn't fulfill us, it sections us off. Work requires the division of the self. What could actually fulfill us is life after work, what we do not working. Fulfill us in ways that humans have been dreaming about for tens of thousands of years.

That work isn't the only thing that makes us divided, incomplete beings, and it's not the first. There are natural processes at work complicated by the fact of living in a society. we become divided, incomplete. Watch a two-year-old react when their uncle who does endless piggybacks arrives at their house, their uncle who will hold them upside down and walk around the house until their face turns a little bit too red, and then you can sit down for a second, wait for people's faces to return to their normal shades, and then it's back up and we can do another lap up- flip side upside down around the living room.

Watch the two-year-old celebrate their uncle's arrival. Watch them expand and crumble with joy. In their faces, yes, but just as much in their elbows and toes and knees and fists. This two-year-old is experiencing life fulsomely with their full body Childhood is the time when delights, joys, pains, and sadnesses are all full body experiences.

Childhood is a time of completion and thus repletion. But adolescence, puberty, this is the body coming apart, separating. This is the beginning of what Freud called the tyranny of the genital function, which is terminology that I will also be using 'cause I think it will get me around YouTube censorship.

The other animals obviously go through a similar transition where they develop the genital function part of the way through their lives, but it doesn't make them neurotic because they're able to crest into physical maturity. They just outgrow childhood. But humans, we have this prolonged childhood, and this is when we develop language and learn the basics of human knowledge and the different ways of thinking.

But it also makes it so that puberty is a wall that we run into. We haven't outgrown the special momentousness that life has during childhood. We've just suddenly hit a deadline. The genital function attains primacy, but we're still haunted by the incompletely forgotten charms of childhood, the fullness of childhood.

This is why Freud calls it the tyranny of the genital function. The rest of the body becomes separate and secondary. It's subordinated. Subordinated when what we really want... Well, there are dozens of words for it: fullness, being at one with yourself, unity, harmony, integration. There's a whole thesaurus entry worth of similar words and related concepts.

Comes up a lot in art and poetry. I don't know how much I have to document this one, but just for example, what is the body electric except for the body that's enthused all over, all at once because it's charged all over, all at once. But it comes out in the religious traditions too, maybe most explicitly in the mystical or heretical traditions, but mysticism and heresy are often just the implicit ideas of the old mainline faith being brought out to the surface.

And there are a lot of political movements. Some of them are cool and good, but, we live in this world, so most of them are wretched. how Eric Hoffer talks, the true believers so often feel themselves to be broken vessels, and they can't fix the crack in the bottom of their carriage.

Maybe it's a hairline crack. They can't even tell where it is. But ideologies fascism promise wholeness. Fascists will overtly talk about the state as an organism, one organism, one body, and then fascists will vow to make that body hygienic and healthy with all the toxins removed. No more impurities, no more foreign entities, and thus a body politic that's at one with itself.

AI and automation didn't cause the division of the self, the division of the body. They exacerbated it, but they didn't inaugurate it. AI and automation didn't cause the division of the self, but they could be what fixes it. If AI were ours in common, if it were our joint worldwide unearned prize for happening to be born in this technological moment, then it could give us the leisure to start to live fully again, to live lives of completion, repletion.

Work can't fulfill us. Play can. The euphoria of writing a sonnet, lingering in the grass, writing sonnets to each other, that's a euphoria of the whole body. The part of my forearm that hurts sometimes when I do a certain pull-up, that's never felt better than when I was writing a poem that I liked.

Or sports. Maybe this one mostly goes without explaining, but especially sports soccer and basketball, where, it's not just about physical recreation. There's so much to do with personal style. It's about how you dribble, how you look at and see the field or the court, your sense of what good pacing and spacing are.

That's all a matter of aesthetics, of personal style. So many of, Allen Iverson's crossovers, they're masterpieces of abstract expressionism. But writing and basketball, those are just the things that I do. There can be, and there are, playfulness in all kinds of things, all kinds of art, all kinds of productive and non-productive activities, things that enliven and exhaust and enliven us.

Oscar Wilde is, the presiding spirit of this essay. He, for one, certainly thought that there was a supreme playfulness to talk, to pronouncing droll aphorisms while holding a glass of golden wine. Blake wrote that exuberance is beauty, and lives without work, lives where we've automated work and we're just off playing, playful lives would be exuberant ones, and thus, they'd be beautiful ones, too.

I don't think that, full automation is imminent, but increasing automation, it could, at the very least, reorient our lives around leisure instead of work, work days. And in the world of increasingly automated luxury earth communism, while the automatons automate work, humans would increasingly come into the full flourishing of our humanity.

Let the robots be efficient, and we can be blithely inefficient. Let them economize so we can indulge. Let them minimize their energy expenditure, move about silently, without dallying, so that we can revel in our uncalculated excess, moving about as noisily as we please, dallying. This is what AI should be giving us, but it isn't, and it won't as long as it's a capitalist enterprise.

The opportunity cost of privatized AI is nothing less than the coherence, the fulfillment of our lives.

We've just heard clips starting with

Political Breakdown tracing how Peter Thiel guided and funded J.D. Vance's career in service of an ideology that sees democracy as an obstacle.

The Tea with Myriam François compared today's AI to a phone's autocomplete, noting it predicts words well but fails badly and understands nothing.

Elena OnOffline examined the battle of AI-aided processing of high-level mathematics, raising questions about credit, integrity, and whether original research is becoming a resource race.

Tech Won't Save Us argued that Musk isn't a libertarian but a capitalist who leaned on federal contracts, loans, and subsidies decades before rivals followed the same path.

Better Offline called out the media for uncritically printing Jacob Coxon's vague, scary warnings about AI, when he never gets specific or holds the companies accountable.

Blood in the Machine explained why worker fights over AI are rarely about AI itself, but about control, oversight, and defending dignity against technology imposed from above.

Adam Conover described how AI scrapers and marketing bots are destroying the independent web, killing small sites and turning Reddit into machines talking to machines, making the dead internet theory come true.

And Nash Keune imagined a world where communally owned AI frees people from work, giving them leisure to live fully, since work divides the self while play makes us whole.

And those were just the top takes, there's lots more in the deeper dives sections,

If you get value out of the show - and want to make sure we can keep going while getting it delivered ad-free to the new, members-only podcast feed that you'll receive, sign up to support the show at bestoftheleft.com/support - there's a link in the show notes - through our Patreon page, or from right inside the Apple Podcasts app.*

If you have a question or would like your comments included in the show you can record a voice message - re-recording until you're happy with it - by tapping the link in the show notes,

As for today's topic,

The big problem with ideology is that they're so often designed to be unfalsifiable from within. As an ideology begins to fail, the response by the defenders is always that their ideology wasn't implemented strongly or purely enough.

When capitalism crashes, its defenders blame regulation and demand even more capitalism. When communism fails, its defenders claim that real communism was never actually tried. It's an easy move, and it's self-sealing.

It's not so different from how every counterpoint to a conspiracy theory can feel like more evidence of a cover-up, logic be damned.

Now, AI has effectively turned that unfalsifiable mechanism of claiming that every failure is only evidence of the need for more of the same into a business model.

The expectation of constantly falling short of the goal and constantly having to invest a little bit more, add a little bit more compute power and tweak how the AI models function is literally the plan. We've already watched it happen. When the old approach of just building bigger models flattened out in late 2024, the answer on the shelf was a new kind of more, reasoning models that think for thousands of words before they answer, and the spending kept going up. The plan is to claim the end goal always requires just a little bit more but there's no natural point to judge progress or to evaluate whether it makes sense to push forward, because pushing forward is the only option they've given themselves.

The economics of scale sort of run in both directions with AI. The traditional understanding is that prices go down as scale increases, and that has been true with AI, but it gets talked about a lot less.

A few years ago, I heard that a simple AI query took something like 10 times the amount of energy as a Google search, but that's no longer the case. The resources used for that same simple query from a couple of years ago has gone down dramatically because of the normal process of scaling, including better, more efficient hardware and finding efficiencies for how AI models work.

We don't talk about how the cost of yesterday's capabilities is actually getting cheaper because the whole business model of AI is about the frontier and pushing it further, which is the expensive part.

The other reason to not be focused on yesterday's capabilities getting cheaper is the cumulative effect explained by Jevons' Paradox. A couple of years ago, comparing an AI query to a Google search was around 10 to 1 but although the cost of a single query has dropped quite a lot, the total number of AI queries has gone up dramatically, as nearly every company has shoehorned AI into their products in one form or another.

Way back in the mid-1800s, when steam engines were the cutting-edge technology of the time, burning coal, it was thought that building a more efficient steam engine could effectively reduce the amount of coal that needed to be burned in aggregate. It turned out that coal use went up because the greater efficiency of the engines meant that they could be used more often and in more places. The coal usage per engine went down, and the cumulative usage went way up. That's Jevons' paradox and it's happening again with AI usage.

So the need for "just a little bit more" is feeling the pressure in multiple directions. The cutting-edge frontier of AI requires multiple times more energy and compute power than the previous generation, and that curve keeps bending upward with each new frontier model. Meanwhile, the long tail of AI usage is expanding the aggregate use of AI simply because it's almost everywhere you look, whether you asked for it or not.

So, infinite growth and infinitely greater power consumption is the plan because there's no off-ramp, no point where the masters of AI would say "enough," which brings us to the financial side.

One of capitalism's superpowers is to innovate ways to separate the risk and costs from those who stand to profit the most. It started honestly enough. A founder of a company could grow faster and reduce their own personal risk by bringing on private investors, and those investors took real risk away from the founder in exchange for a real piece of the company. From there, it expanded to public markets and regular investors who could buy in through a stock market.

From there, they went to the government and figured out multiple ways to extract what amounts to investment from taxpayers, with one difference. The taxpayers took on the risk without getting the piece of the company. Sometimes it takes the shape of direct investment and loan guarantees, and other times it looks like tax incentives and exemptions. All of this has led to where we are now, which is that the AI companies are effectively finding a way for the public to bear some of their burden in the form of increased prices to ratepayers for the water and electricity cost that everyone needs to pay.

And for AI, this dynamic also runs in two directions at once because they're not just drawing down local water supplies or using power from CO2-belching power plants, driving the climate crisis, they also trained their models on the cumulative output of all human knowledge. An abridged and biased version of the cumulative output of all human knowledge, but that's what they'd like.

Banks getting the government to foot the bill when they crash is one thing but vacuuming up all human knowledge is the greatest version of socializing the costs and privatizing the gains in history.

Just so there's no confusion about the progressive perspective on public investment to help reduce risk for big things, we're generally in favor of that. The problem is when the public bears the risk without gaining equity or influence. History has shown that when the public invests in disruptive change without implementing proper controls, oversight, or public ownership, what you end up with is the Gilded Age.

So, in a world of ideologies that mark any shortfall as evidence that we need just a little bit more of the same, what we really need a little bit more of is democracy and public ownership. It's a lesson we could have learned over and over again from examples in the past .Railroad companies were granted land owned by the public, and the private ownership of those rail lines helped create the robber barons, all while those steam engines were the Industrial Revolution running on public land, burning the coal that seeded the climate crisis we're experiencing today.

But the pattern has never been more clear than with AI. The source of their value came from the public knowledge, they would be nothing without it, they are knowingly externalizing their financial and environmental costs onto the public as well, and the promise implicitly made by their business model is that they will never stop. There will always be a little bit more to do, a little bit further to push, and there's no internal braking system in their ideology that could steer them to do anything different.

An ideology responds to setbacks by doubling down, whereas democracy is a procedure rather than a vision and it's built to manage and resolve problems for society when they inevitably arise. Democracy gives agency to those impacted and, ideally, accountability for those who cause the problems. Public ownership is the same idea. Publicly or cooperatively-owned companies can choose to act for reasons other than profit, including slowing down or stopping, which the private owners of AI have designed out of their plan.

All of this is the spirit of what Peter Thiel wrote back in 2009 saying, "I no longer believe that freedom and democracy are compatible." His version of freedom is to never be accountable to those whose lives his businesses are impacting, which is why he and the rest of the tech bros who either started on the right or have taken a hard right turn in recent years see government as not much more than an obstacle for them to overcome. And they're right, it is an obstacle to them because democracy is there to work on behalf of the people so when corporations and the wellbeing of people are at odds, it is the role of government to be an obstacle for the benefit of people.

Trump opened the door to public ownership when he started having the government take stakes in private companies, in Intel's case without any influence over it at all. The left should take that precedent and run with it but with the addition of demanding representatives of labor and the public interest be placed on the boards of those companies. And yes, a public stake is only as good as the democracy holding it, so there's work to be done on multiple levels. The public ownership and the functional democracy have to come together.

In the meantime, there's also worthwhile regulation that is being tried out. Ohio regulators now make large data centers sign twelve-year contracts and pay for at least 85 percent of the power they reserve whether they use it or not, and Virginia has a similar rule. This means the companies are shouldering the risk of the power they asked for so if they bail on their plans, the community doesn't get stuck with the bill.

Capitalism got great at spreading the risk while concentrating both the profits and the decision-making. But decisions concentrated among a small elite class of the ultra-wealthy have no accountability mechanism, precisely as they like it. So, when something as potentially disruptive to everyone as AI comes along, there's no good argument for leaving decisions about it in a very few hands. The stakes are too high to leave democracy out of the loop.

Demanding public ownership is about more than heading off the next gilded age or, at least, preventing the one we're already in from getting worse. It's also about the much-needed oversight of this technology that is full of possibility, both hugely positive and mind-bogglingly negative. Guiding decisions like that is what political democracy and democratic ownership is for.

And now, we'll continue to dive deeper on 4 topics today. First up;

Section A, The Nerd Reich - Who They Are and What They Believe

Followed by Section B, What AI Is Actually Doing To Us

Section C, The Grift, The Bubble, and The Doom Marketing

And Section D, Fighting Back - Workers, Communities, and Alternatives

There's a long history of people seeking to escape democratic societies or else to bring back monarchy or authoritarianism. In 1997, there was a book that came out called The Sovereign Individual. It argued that in the 21st century, two technologies would result basically in the overthrow of democracy and the demise of nations like the United States: cyber currency, which we would call crypto today, and the other one was advanced automation that would create a world without jobs, and that wealthy individuals who are savvy investors could profit from this coming doom and prepare for it by making the right investments.

This book had a major impact on Peter Thiel. 

And of course, it influenced the people Thiel was paying. Joe Lonsdale got to work. He started a nonprofit, California Common Sense, allegedly dedicated to transparency in government spending so that voters could make a difference if they found things they were unhappy with.

On the surface, that sounds great, but two founders of that nonprofit started a for-profit, OpenGov. OpenGov is a software service for government agencies to manage data. The idea of replacing democratic processes in the hands of all of us with technology in the hands of few was becoming less fringe, and other nonprofits with similar goals started popping up.

But while the idea of 

using technology to make a decentralized country had some illustrious backers in the mid-2010s 

I'm going to leverage technology to create a new country. And yes, I did say country, not company ... 

it never really materialized on a serious level. But the think tanks would herald a few successful versions of their privatization plan.

Like Sandy Springs, Georgia, famously the first city in America to privatize nearly all government services. It was the darling of the libertarian press. Everything for profit, nothing public, and a consulting firm making a little cash off the taxpayers at every juncture. Those extra taxes for corporate profit started adding up, though.

Eventually, Sandy Springs un-privatized, saving taxpayers millions of dollars a year. 

There are also nascent communities focused on Christian values that the network state media heralds. 

Battle Ground, Washington, where a data center billionaire is buying up all the property to own the town, the mayor is on his payroll, and Highland Rim, a series of developments in Appalachia run by the investment firm New Founding.

And that's not even including the California Forever project, a proposed tech city in Solano County funded by Marc Andreessen, Reid Hoffman, Sam Altman, and other billionaires. But what if this was never really about these little mini-governments? Here's Peter Thiel saying the quiet part out loud in 2010.

Technology is this incredible alternative to politics, and the task in this world where politics has become so broken and so dysfunctional is to find a way to escape from it. 

Tech instead of politics. We already had Doge and Palantir becoming the operating system of the United States, with a stated goal of being more powerful than the governments using it.

How can we help the government stop the bad guys while also putting systems in place to make sure n- nobody within the government was doing anything inappropriate? So you have to have p- you have to have systems that watch the watchers. 

After leaving Palantir, Lonsdale, through his firm Eight VC, continued investments in what's called GovTech, tech for government clients.

That's surveillance tech, AI drones, AI weapons, even tech to replace 911 with AI. All aim to replace, privatize, and profit from tasks normally attributed to democratically accountable officials. 

What's happening now, though, is that the companies are becoming more powerful than the governments they serve.

We suddenly have all these billionaire oligarchs spending so much money, they were able to actually capture the federal government with Trump. So there- it's this interesting moment now where we're getting a glimpse of what government by technology company would look like. 

And Lonsdale is working to convince you that we need this by making you scared.

Lonsdale founded the Cicero Institute, a think tank that pushes anti-homeless messaging. Not anti-homelessness messaging, but messaging attacking the homeless themselves, and even legislation creating concentration camps where unhoused people would do forced labor. 

You actually can go into states, you can, partner with governors, you can pass laws.

Cicero's in 15 states, we're getting dozens of laws passed. And, and then we're gonna teach people that and go to the national level. 

Riling people up is definitely part of their game plan. That's what we saw in the San Francisco recalls. San Francisco had a spike in crime during the pandemic, but so did almost every big city in the United States.

Only in San Francisco, which has extremely low crime rates compared to most cities in red states, did they create this massive panic using social media and doom influencers to convince everybody that everything was worse than it'd ever been. 

And as Lonsdale got into nonprofits and think tanks, Patri Friedman, remember him, Milton Friedman's grandson, started his own VC firm with backing from Peter Thiel called Pronomos Capital, existing entirely for investing in privatized governments.

Próspera is this little enclave on the island of Roatán in Honduras that claims to be its own little charter city that was funded by American venture capitalists and has been touted as the flagship project of the network state movement.

But the real flagship project is something bigger, the entire United States government. It shows in policy. Remember the whole thing about taking over Greenland? That idea emerged and was celebrated by the people we've been talking about. 

I think having a frontier is very healthy. I think part of America being great is that it's, is growing and, and creating more wealth for everyone.

If the people in Greenland wanna be part of us, if we can use those resources better and make sure Russia and China don't get advantages in the Arctic. 

The freedom city as a name didn't really gain momentum, but the idea is there, and the influence in the administration is clear. Joe Lonsdale helped staff up the Trump administration, and just recently accompanied administration officials to the Philippines to christen a new regulatory zone.

Just last week, we signed a historic deal with the Philippines, 4,000 acres right next to Subic Bay, where we're gonna build a first of its kind AI native industrial park. 

But I was there for a day in the Philippines with, with everyone. they're, they're excited what you guys are doing. 

That sounds a lot like a startup city.

They've also embedded themselves with the rest of the right-wing movement. 

I could not have envisioned that JD Vance, a Peter Thiel protege, would be the vice president of the United States. And so when Vance got onto the ticket, then it became real that this stuff could travel to the White House. In September 2024, there was a conference at Fort Mason in San Francisco called Reboot 2024, and its website announced that we were now in a new reality and that Silicon Valley was waking up and about to become this powerful force in politics.

And the guest speaker was Kevin Roberts of the Heritage Foundation, who is the guy who had shepherded Project 2025. So here you had tech VCs in San Francisco and Kevin Roberts in the same venue proclaiming the new reality, and the new reality is where we are today. 

this tech industry that's so high on its own, supply that they-- that there's this belief in these, fantastical sounding ideas such as, as, you get into in, in your reporting and, the singularity, right? And that all of these notions around building an intelligent, machine are rooted deeply in, as you said in the intro, eugenics, the idea of measurable intelligence, and then also these ideas that fall off of that world, the way of thinking, which is this very hierarchical approach to, life and, ultimately an extractive, exploitive, worldview that a- as we see today playing out in the AI industry is, alive and well.

They're not exactly subtle about it, are they? 

No. 

They're, it was interesting with the film. how did you realize what... How did you get to the thesis of your film?

That's a great question. I don't think that I had a thesis going into it. I genuinely wanted to know, what is artificial intelligence? Who's building it, and why? And I think when I was asking those questions initially, I really did believe there was such a thing as artificial intelligence.

I just hadn't even questioned it. I was well, of course, And in discovering that, oh my gosh, there's no such thing as artificial intelligence, it's a construct that is a marketing term, and when we hear it, we need to be extra careful to examine, what are we being sold? And the, the way that race science plays into the contemporary dialogue was, was actually how I first sniffed this out.

It was, it was, 

contemporary groups who were concerned with AI safety, which, for the- Our friends 

the rationalists ... 

our friends the rationalists. And, and, and again, AI safety, I was "Oh, that's cool. Yeah. Oh, oh, AI alignment. Yeah, we want to align these computer systems to comp- to human values."

And so, the moment that you start to read stuff from these people, it is clear the overt references to race science. It's not even, as you say, a, a Nobody's being shy about it. And there's a podcast, I didn't put it in the film 'cause it was just too, crazy, but there was this podcast, interview with Lex Fridman and Sam Altman, and Lex Fridman asks Sam Altman, "Hey, you're building a system that's gonna know all of the truth of the world, right?

it's gonna be this all-knowing system. How are you gonna confront the fact that, it might give answers that are truthful but most people aren't ready to hear, such as, theories of intelligence and race?" 

Oh, so Lex Fridman went there? 

Yes! And he go- he did a whole podcast on the bell curve.

he, he, he goes there. Yeah, and he was actually the first person that I heard, openly entertaining ideas that I knew just from being a person in the world, were problematic. And I say problematic not to be thought police. I say problematic because they are ideas that are factually incorrect, but, but deployed in the, subjugation of others, right?

Or the justification of the subjugation of others. So, when you, when you hear those ideas, that you're "Wait, put the brakes on. That's a stupid idea." So that was the first time I was wow, there's not even a shyness around entertaining these really dumb conversations, and then you dig one level deeper, and that's when I found some of your work.

And I know that when we first chatted you were the... You, actually pointed me in the direction to tons of information that's not in the film, that could be a whole, should be a whole other film, how a lot of the current, conversations around rationalism and AI safety, if you go back to 2010, are, are rooted in, declarations around, racial hierarchies and intelligence, which is just shocking.

1990s, that was when Nick Bostrom, the, intellectual heav- the intellectual heavyweight of the LessWrong sphere back in the day was posting his, straight-up racism and dropping N-words and, that was the one he apologized for and got in trouble for. He didn't apologize for the ideas. He still thinks Black people are stupid, but he apologized for using the word 'cause racism is when you use bad words.

He-- these, a lot of folks who, play in this hierarchical way of thinking about humanity will employ a double speak, so they are not unable to be held to account, right? So what Nick Bostrom said in his apology was it wasn't his field. He is not somebody who's an expert on race and intelligence and therefore couldn't weigh in 

That didn't stop him saying it 

That's insane.

It's a, it's a lack of accountability for these really harmful ideas that is a, a definitely also a through line through, this way of thinking. And it's important to call ... I really, this is why I really appreciate the work that you do, David, because it's really important to call this out.

Sorry, this, these, this, these things out. 

That's okay. We only have to bleep it for Apple. But, the, the good thing is he did actually get a small consequence. That is Oxford University finally went, "What on earth is going on here?" And shut the Future of Humanity Institute down 

I gave, we did a screening of the film in Vienna in June, July, June.

And, one of my friends who goes to school in Vienna was "Oh yeah, no, in my class this semester we read a book by Nick Bostrom." 

Superintelligence. 

It wasn't Superintelligence, it was another one of his books. I don't know, whatever. And I was "Whoa." So i- I was "I thought he was canceled" and so I was very excited.

I was "Oh, you need to come to the film and you need to bring your friends." But that he's still finding audience. I think that it's, the, the, the people who espouse these ideas- I 

mean, there are still academics 

who take Nick 

Lane seriously, 

100%. And and, and if we look at what's happened, this is a slight abstraction but not really at all.

You look what happened with, Jason Arday in, at Cambridge. there's, there's a the, the way that the system perpetuates a certain narrative around who belongs in academia, is, is insane. And it, and it's, and it's very real and active and alive. And we see a really tragic consequence in that particular story.

but- 

Where he was badgered to death by a bunch of literal straight up documented race scientists. 

Yes. 

And- 

Yes. the, it s- yes. And so, th- this isn't, this isn't a thing of the past. This is a thing that is actively, exists and is also actively becoming normalized, i- further into the mainstream in ways that I'm, super uncomfortable with, as should everyone be super uncomfortable with.

So when you realize that these people espousing these, ideas about so-called artificial intelligence, have problematic, worldviews and also a lot of... A- a- and problematic hearts. I think that's what gets me the most about all of this, is it's what person wants to believe that we should be building at the expense of the planet and at the expense of copyright law, That's 

a very, that's a very pol- that's the politest way I've heard today to say huge assholes.

Yeah, they really are. And, that, so that realization that these people are perpetuating their story, that drove me to make this film. And I'm so, feel really blessed that, you and so many other amazing folks, took the time and the risk to chat with me through, a Zoom like this, and trust that it would turn into something that would be a film.

And then I am genuinely, I don't know what I was expecting, but I am so pleasantly surprised at how, this film is resonating. 

I've always, I've always realized this whole field of AI was full of massive race scientists, as I kept colliding with them. Yudkowsky was pushing straight up race science in 2006, 2007, and somehow it never registered with them.

And I went, I was going to some of the London Less Wrong meetups for a while, 'cause I hung out on the site for a few years, 'cause it's an amateur philosophy site. Arguing philosophy is fun. And, so I went along, and I think last one I went to, one guy went in this massive race science rant, and I went, "This isn't a great area."

And I didn't show up for any more. I stayed on the site though, but yeah, for a while. But, yeah, it's absolutely bizarre people. And then you realize, oh, this is just the eugenics guys who we've always had. 

as soon as Jacob tweeted this out, a bunch of other tech workers and AI workers at Anthropic, still employed in Anthropic, said, "We totally agree.

We also believe that, there's a 10% chance that AI could kill everybody." But... And it's a tough, it's a tough problem, and we're gonna keep building it regardless. We are gonna build it 'cause we know better, and we are making this decision, again, atop the hierarchy. And so my question is, what, what do you make of this?

'Cause viewed from a certain light, this seems like an awfully fascist viewpoint. We know better. We will put the world at great harm, it's not mentioned in the tweets, for personal profit and power, and we will make the decisions about how it rolls out. 

I think what we're seeing in their willingness, according to their own logic, the- their willingness to risk human civilization is a prelude of the authoritarian infrastructure they want to build, right?

these guys say that AI is the final technology, the ultimate, invention. So they talk about it in this, millenarian way. And in, in a world where you are the last company standing, you have total control, right? Where you've, you... So they're imagining a future where they don't have to listen to the plebs anyway, so they're just doing it now.

I, I also wanna say that that 10% figure is getting a lot of traction, that he's saying, "We think there's a 10% chance that we'll destroy humanity." Dario Amodei has put that, the odds at 25%. So it's one in four. And, I think you can, we can debate whether that is an accurate wager, or whether their tools are capable of this.

What I believe is that what that tells us is 25% is an acceptable number for Dario Amodei, right? Just like I think Elon Musk said there's a 20% chance that this technology will annihilate all humans. He thinks that's an acceptable number, right? Can you imagine saying that with a straight face? so they- These, these, these moments, these outbursts, right?

These incidences, they're called, tell us a lot about the people behind them. I think the passage you read is, is really important because, again, we're... Our analysis is ultimately material. It's there are these dangers in our economic, political, technological systems, and there are consequences unfolding because of decades of inaction.

And so we are entering a world where these people are saying there's gonna be crises galore, right? M- there's gonna be more and more inequality 'cause we're building companies that, whether or not they deliver anything useful, are going to siphon up power and hoover up wealth. That means all of you are gonna be poorer and subjected to more and more disasters, and we're gonna decimate public health and all the like.

And, and they need an ideology that rationalizes that, and the ideology that rationalizes that inhumanity is fascism, right? So there's, It's, it's not surprising that this is where they've- That this is, his name said when you go down the road, you go, you wind up where you're going, because there's really no other way to justify, justify this.

I'm... I think the, the, this, the virality of this bit of news is interesting because I, I think, it had some- I think it had something 100 million impressions last I looked, something remarkable. And while I am not in the same ideological camp as the AI safety community, I'm not mad that this guy is spreading the message that these guys are willing to kill us all because, I think they're willing to kill us all in lots of other ways, like by burning tons of carbon and by just, again, sucking up wealth so we're destroying the welfare state, which has enormous consequences for people's lives and, it, it, and their willingness to steal people's jobs.

so they're, they're, I think the story he's telling can't be the only story that's out there, but I don't think it's necessarily a bad thing to have circulating in the media if it's an opportunity for us to complicate the narrative a bit. 

Yeah. Yeah, and I, I think, and the only thing I would add there is that I d- I do believe that the only apocalypse that the, people driving the corporate AI arms race are, really genuinely engage with is the apocalypse of them losing the AI arms race.

And I do believe they believe that to be genuinely apocalyptic. the loss of their wealth would be genuinely apocalyptic. Mm-hmm. and we are already seeing the way this technology is eating the internet, right? and, the remaining media organizations are feeling that so intensely right now because people are becoming increasingly reliant o- on, on, on these tools that they're being driven to.

as we all-- as as you cover, it's getting pretty hard to do just a, a Google search, right? so, so that's the apocalypse that they are, are concerned about. It underlies why they're constantly re- calling one another evil, right? And, and, and this story of it has to be me, it has to be me because I'm the only good one left, right?

that, that, that goes back to Musk and g- s- and Altman saying, "It's Google that's evil," and then Musk saying, "It's Altman that's evil," and then it's g- and now it, it, it's, it's, it's Anthropic saying OpenAI, Google and, and, and, and, and, and xAI are, are evil, and then all of them just leasing, compute from Elon Musk.

but at any rate, They, that, that is the apocalypse that they're, that they're worried about. The rest of it, it, I think is just, I, I think the Earth has become a non-player character in their, in their narrative to, to, it, and not, not just individuals. It's just anything other than that. And you can see that a, that, that sense of apocalyptic panic at the loss of wealth In the way that they respond to, a 5% one-time billionaire's tax, right?

It truly is the end of the world if you're David Sacks. it's the only thing he gets excited about, in terms of, of, a genuinely bad outcome. So, this is one of, one of the conclusions we come to in the book is that, is that this extreme wealth is deranging and is a, a WMD in it, in and of itself, and we really do have to get at the core of it.

And it can't just be a slogan of, every billionaire is a policy failure. we really do need to abolish billionaires, because they are a material threat to the rest of us, and, before they all become trillionaires. but the other argument that we make in the book, and that's really key, is that this, this Wild West frontier that they want, where it is possible to just speculate that maybe you'll...

You have a 25% chance of wiping out all of humanity, and the government is twiddling its thumbs and acting like it can't do anything, is why they created the alliance that they did with Trump. we, we are making the argument that if you wanna know why Silicon Valley aligned with Donald Trump, lined up behind him, it has way less to do with the psychology of them being upset about their Marxist kids, or, the, some slight of Joe Biden than the fact that they wanted exactly what they got, which was, which is absolutely no regulation.

And, I don't wanna fetishize what the, the, the, the f- pretty timid regulations that the Biden administration was putting, up for, for, to, to regulate AI. But they were trying to, to, to, to take seriously these existential risks, and do things like label deepfakes, and, say, "Okay, if you think it's a WMD, we should treat it like one."

and that is what led, Marc Andreessen et al to have a giant temper tantrum and decide to line up behind Donald Trump. So the real issue is, is, is not are they right or are they not? You shouldn't be allowed to build a technology that you don't understand that might do these things.

There should be consequences for this. there should be... I th- Rochona responded th- th- they, they should be held legally responsible for the harms of their models. and I think that if we were taking-- if we had taken that seriously a long time ago, we would not be where we are. and it's the fact that they got their frontier, that there are no consequences, that they have i- seem to have immunity for the co- for, for the outcomes.

look, I live in British Columbia. We had a mass shooting here, a year ago. and- Some of the families are suing OpenAI because it seems to have been-- It was a school shooting. It seems to have been done in collaboration with a chatbot. and OpenAI seems to have known this enough that they took away the account of, of, of the shooter because they were so alarmed by what they were seeing.

And then he-- Then, and then, and then they created another account. and, and what we got from Sam Altman was an apology. 

Yeah. Yeah. 

he said he was sorry. 

Yeah. Oops. 

how is that possible that we got an apology for a mass shooting, for a mass casualty event that was entirely predictable, and they're just going on.

How, how is it that there's no accountability for, for, for, for, for war crimes that we're seeing? So it's just gonna escalate unless there is accountability. And as soon as there-- as soon as the buck stops and there's leg- legal repercussions, criminal repercussions, then I, I don't think we will be having the conversations that, that we're having right now.

AI critics like Karen Hao have explicitly compared the logic of Silicon Valley to that of empire and colonization 

Data is the last frontier of colonization. They took our land and then tried to sell it back to us. Now they're taking our data and trying to sell it back to us as a service. This idea of just perpetuating the colonial dynamics.

Wealthy countries get to come into poorer countries and do whatever they want, extract whatever resources they want. Not just data, right? Also Critical minerals and all of the things that you need to build data centers 

These guys are happy to take advantage of exploited labor around the world while pillaging natural resources from places that won't ever see the benefits of this glorious tech, because again, humans, we're just fuel.

But they're also colonizing our broader humanity. they're extracting attention and value from us even when we're unaware that it's happening. In doing all of this, the logic of Silicon Valley is not simply anti-human, it's actively creating a project of dehumanization, one that bears some similarity to the active dehumanization necessary for strategies like colonization.

During the era of European colonization, a similar story about rationality and enlightened human civilization was used as justification for the dehumanization of a lot of people for the sake of the prosperity of a very small group of people. In this system, the spread of rational civilization was used as a mask for a ruthless desire for power and wealth.

To gain these things for themselves, they needed a logic that could justify the dehumanization of others. So by turning their specific ideas of reason and religion into universal concepts, they believed that indigenous populations with different concepts of reason or religion were less than human, which both allowed them to ruthlessly exploit these populations while still being able to think of themselves as good, enlightened humanists, and in some cases, as good Christians.

Philosopher and psychiatrist Frantz Fanon argued that the horrors of colonization don't simply dehumanize the colonized individual, they actively destroy the humanity of the colonizer as well. In his book, Black Skin, White Masks, he talks about how the white colonizer becomes a slave to their own superiority, as by considering themselves superior to the colonized, they view those humans as objects, which then alienates the colonizer from their humanity, and from humanity more broadly, making genuine human connection impossible.

maybe this is why in this new era of tech colonizers, all these guys seem sociopathic and incapable of basic human empathy, even making arguments about how empathy is bad. 

Is suicidal empathy. 

Mm-hmm. 

the, it, it, there's there's so much empathy that you actually suicide yourself. 

Yeah. 

so that w- we've got civilizational suicidal empathy going on.

Following Fanon's analysis, it's not that the Silicon Valley guys are dehumanizing society while maintaining their own superior humanity. Instead, they are actively dehumanizing themselves, which then subsequently makes their own thinking, their own political projects increasingly inhuman. And I already said, if they're right about the future of their technologies, they're going to eventually give their own humanity over to the machines that they've created.

Writer and theorist Aimé Césaire describes the fundamental lo- Lie at the heart of the colonial project in a way that parallels the dehumanizing logic of Silicon Valley, while also showing how neither is actually concerned with the project of expanding and improving human civilization. He writes that, "The essential thing here is to see clearly, to think clearly, that is dangerously, and answer clearly the innocent first question, what fundamentally is colonization?

To agree on what it is not, neither evangelization, nor philanthropic enterprise, nor a desire to push back the frontiers of ignorance, disease, and tyranny, nor a project undertaken for the greater glory of God, nor an attempt to extend the rule of law. To admit once and for all, without flinching at the consequences, that the decisive actors here are the adventurer and the pirate, the wholesale grocer and the merchant, appetite and force, and behind them baleful projected shadow of a form of civilization, which at a certain point in history finds itself obliged, for internal reasons, to extend to a world scale the competition of its antagonistic economies."

We could paraphrase Césaire here and say that we need to first agree what the project of Silicon Valley is not. Neither human liberation, nor philanthropic enterprise, nor a desire to push back the frontiers of ignorance, disease, and tyranny, nor a project undertaken for the greater glory of human thought, nor an attempt to make the world more free and democratic.

Following this, we can admit once and for all that the decisive actors in the tech authoritarian age are digital pirates, colonizers of attention, exploiters of human labor, all motivated by little more than to push the limits of capitalist exploitation while eroding the counter effects of democratic politics.

In other words, fuck humans, get money. A recent study by The Lancet looked at the effects of Elon Musk's cuts to programs like USAID, which he enacted while in charge of the Department of Government Efficiency. These cuts, allegedly fueled by the clean code of logical efficiency, dismantled a program which The Lancet estimated prevented ninety-two million deaths globally between 2001 and 2021.

It then projected that these cuts could lead to an additional fourteen million deaths between now and 2023. so in less, in less than four years from right now. Musk and those like him will sell the public a tale of human advancement, radical efficiency, utopian possibility, when the reality is dirty drinking water, large-scale exploitation, and 14 million dead bodies, with at least a third of them predicted to be children.

But while using colonization as a framework to understand Silicon Valley dehumanization in the age of AI is helpful when we're trying to understand the scope of its exploitation, treating this project as just the contemporary version of colonization misses out on some of the really important differences between the two, including those which highlight our own complicity in this project, as well as our radical potential to fight back and reclaim some of our own humanity.

Quinn Slobodian writes about this in a recent essay in which he argues that invocations of empire and colonialism also risk being trapped in the register of polemic rather than analysis, which is a crucial distinction to make in light of the fact that many of these discussions are happening on platforms operating on Silicon Valley logic, which prioritize polemic and disincentivize more careful analysis and critique like this one, like this platform we're on right now, like YouTube, right?

I would say it prioritizes polemic and incentivizes more careful analysis and critique. So everyone watch this video to, to prove them wrong. By watching this video and sharing it with your friends, you're technically, you're technically fighting back. He writes that for those living under direct colonialism, subordination is very difficult to ignore.

The harvesting of personal data is different. It is a form of oppression that remains largely invisible to those it affects, and that we often need a third party to reveal to us. To make the analogy with colonialism hold, one would need to invoke the contested notion of internalized oppression. it's not just that our experience of being oppressed by Silicon Valley is wildly different from what happened to the victims of colonization.

Because while enslaved peoples weren't reaping the benefits of cocoa sales in Spain, many of us are financially benefiting from the growth of things like publicly traded AI companies. Slobodian writes about our entanglement with big tech stocks, noting that more than half of the American population is plugged into them, whether directly through stock ownership or indirectly through pensions, mutual funds, and retirement accounts.

And listen, I know most of you would never dream of purchasing stock in ChatGPT or SpaceX, but if you have a 401or a pension fund, a retirement fund, if you invest in an index fund, well then you are absolutely set up to benefit financially from this project of tech colonization. Which leads Slobodian to call for us to think of ourselves less as digital natives and more as digital Creoles, with one foot in the world of human community and the other in the world of digital exploitation.

Because the apps and devices might make us feel like shit, but they're also maybe our only hope of retirement or home ownership in an economy without any serious social safety nets. In other words, we need to recognize that unlike the colonized person, we co-create value in our interaction with networked technologies.

In this way, we might be undermining ourselves when we use the framework of colonization to understand the situation, rather than thinking of ourselves as workers with collective stakes and collective power. Slobodian references a forthcoming book which notes that the tech workers movement needs to abandon the mystifying language of serfdom and empire for the practical terminology of walkout and strike.

Now, I don't think that any of this means that we can't rely on examples that consider Silicon Valley through the framework of colonization, especially when they are mobilizing resistance. But we can't stop there, as while the situations are similar, they are not the same. And because of this, we need to analyze the contemporary situation to best fight against our own era of tech dehumanization.

Another crucial aspect of recognizing that, to use Slobodian's analysis, we are more Creole than we are colonized, is the fact that even if we hate data centers and we use dumb phones, we're still internalizing the logic and values of this system. To speak more specifically to the topic of this video, we are internalizing a logic that is perpetuating our own dehumanization because the more we're thinking like algorithms and financial models, the less we're actively becoming human.

the question is where does government fit into this picture? you quoted President Biden's last speech warning about the tech bro, industry complex, the technology industry complex, or TIC. but what are Democrats doing right now in Congress, and what are Republicans doing?

Largely not going after the Silicon Valley oligarchs who are behind Trump. a lot of people are attacking Trump and his- and criticizing his family, criticizing Vance, but we don't see a lot of direct criticism of these billionaires who are enabling the entire thing. The way I look at it, Donald Trump and his- is being paid billions of dollars to do these things, right?

When's the last time we had a president make billions of dollars while in office direct... This is direct transfer, including from foreign governments that are buying into and investing in his companies to curry favor. I think we need to see Democrats speaking directly against billionaire power, against tech billionaire power, against the crypto takeover of our country.

You see all these Democrats trying to say, "Oh, AI, innovation, Silicon Valley innovation." Those are outdated talking points. Silicon Valley is no longer about innovation. Silicon Valley is about fascism and a race to control the future, and that future will not be good for most people, and it's the job of our elected officials, especially on the Democratic side of things, to stand up for the people.

There are a few people doing it. You see Bernie Sanders, James Talarico, Jasmine Crockett, and others who have spoken directly to this, Chris Murphy, Ron Wyden, but people like Gavin Newsom and others are suspiciously avoiding any critique of Silicon Valley, and I don't think we can have that going forward.

People are waking up to this threat. We see it all across the country, and we need politicians who match the energy of the people, or we need those people to be the new politicians. I 

mean, we're seeing this massive movement against data centers, right? Across the country, people are opposing data centers.

What's happening in Silicon Valley? Is there some resistance to this massive takeover and this ideology? 

A little bit. Nothing too noteworthy. Most of the resistance is coming from communities that are being targeted by data centers, that are being targeted by Palantir surveillance and fascism aided by ICE.

You see people resisting and rising up. There are also movements like Purge Palantir, which is boycotting any company that has Palantir products, raising awareness of the role of this company in our government, and putting a lot of pressure on the company. Same thing is happening to Flock Surveillance, where people are starting to militate against putting these big brother facial recognition, license plate recognition cameras in every aspect of our lives.

So you see people waking up to that, and it's not just Democrats, by the way. It's also Republicans and independents, and it's also people who are not activists. It's people who are just becoming activated by the threat of this alien invasion of our country by these surveillance-mad, power-hungry billionaires, and it's pretty inspiring to watch.

It's the grassroots, and I think the politics are gonna have to catch up with that. 

as we begin to wrap up, what is missing from the whole debate around artificial intelligence? 

The question is why we are pursuing this idea that we have this dangerous technology that is going to kill all of our jobs and potentially wipe out all of humanity.

This is the marketing of the AI industry itself and why is that being made to seem inevitable? I don't think it's inevitable at all. I think we have the right to reject this dystopian future that's being imposed on us. How does it make sense to destroy our economy to enrich a handful of oligarchs who are already wealthy and whose wealth has driven them insane, and hasn't made them very happy?

we have this problem where it's becoming a, a competition between democracy and capitalism, and which one is more important. even before the tech bros came along we were destroying the planet for a handful of fossil fuel companies. How does that make any sense? There has to be a different way to do this.

And the same thing with AI. AI is being marketed as this destructive, terrifying force, and we all have to invest in it and adopt it or else. And I don't think it has to be that way, and I think people need to push back, harder on a policy level and stop trying to ingratiate themselves to the AI moguls, which is largely what's happening now.

Even the, the best politicians among us are playing along with this idea that this is an intelligence, which it's not, and that this is inevitable, which it's not. And so I think we need to backtrack a bit and get off of this idea, and question the, the narrative that we have no choice except to bow down to this AI god these guys claim to be building.

It, it's, it's a religion. It's a myth. It, it's not real. I don't think LLMs are completely useless. I think there'll be some good applications for AI. Explain 

what LLMs are. 

Large language models like Claude, et cetera. Mm. I don't think these things have to be useless. I just don't understand why they have to eat everything.

They don't. This is, this is the marketing for the investors. This is convincing people they have to spend trillions of dollars on this so we can create this massive thing that will then suck all of the value out of the economy and render every worker useless. This is absurd, it's ridiculous, and it's an apocalyptic cult.

 

you've been spending time reading the statements and writings of all of these figures. What scares you the most? 

What scares me is that you can be as anti-American and anti-democratic as Peter Thiel and still be a government contractor with investment from the CIA. Palantir was in, was created with investment from the CIA, and I worked in government for many years for Democrats.

I worked for Senator Feinstein when she was chair of the Intelligence Committee, and it boggles my mind that you can s- say that stuff and believe these radical things and still be that deep in our intelligence infrastructure. I have a lot of questions about what went wrong there. to get into government, you have to pretend like you never smoked marijuana, and here you are smoking the most anti-American, anti-democratic dope in the world and becoming a billionaire off of government contracts.

That's what scares me. How did no one catch this? Trying to talk about what's happening to our country without talking about these tech fascists is like trying to explain 9/11 without talking about Al-Qaeda. This cult of dangerous billionaires has a very pronounced idea of what our future is, and it's not gonna have democracy in it, and I don't understand how they got this far this fast.

Nobody... And even today, most people in the media and the political establishment aren't really addressing it

Next, Section B, What AI Is Actually Doing To Us

If your job got stolen by an AI, I guess you have like infinite time to ride a bike, but you also might starve, so that feels bad, right? If there's a data center that is being built near where you live, and now electricity prices have quadrupled, and they're only going to keep going up and up and up, and people care more about data centers than they care about people.

Just like the environmental degradation. It might be really weird to go ride a bike and enjoy that when your child is addicted to talking to their AI companion chatbot. Why would you deal with the, the possibility of rejection when there is a companion that is always telling you how great you are, like on your phone or like maybe in your ear?

This might be coming soon, not decades away. Like this might be happening soon. Trying to build super intelligence actually threatens all of us. This is, this is the thing that like Hinton says, is like, "Hey, it's either gonna be climate change or AI, and I think it's gonna be AI."

So what do we do now? I don't know. I don't know the answer, which I don't know if anybody was expecting that I would, but I will say that there are things that we can all do that will make finding that answer a lot easier. Self-education in a way that makes you critically conscious of power structures.

So I asked Peter for a reading list. If someone wanted to get closer to an answer, what are the educational steps somebody could take if they wanted to help? This is what he recommended. Take a screenshot, enjoy. Which also brings us around to answering my first question, AI, what could go right?

Personally, I think it is the galvanizing effect that it's having in the opposite direction, people realizing what's at stake. And when I asked Peter what could go right, that was what he felt too. 

These companies are currently actually threatening everything, and, I actually do mean that 

How can, how can we threaten them?

This is my last question. What, what does the individual do? How, how do you restore empowerment to someone watching this who is now feeling your doom, but without your, but, but without your vocation? 

there's 8 billion of us 

Go on 

The really nice thing about this is that this is an issue that I've seen unite everyone from the political spectrum from left to right.

Reaching out across party lines and also geographical lines and like ideological lines and people I think fundamentally just want to keep living and lead good lives. 

Same as you and Good Cop. I think what does actually work well is shame and social dynamics. And I think that that as public pressure, the self-interest of 

people not wanting to become pariahs is more motivating than most people give it credit for.

It's funny, this point that I'm making here, that if somebody seems like they're above the law, beyond government regulation and rich beyond comprehension, they're used to their power working. But all the money, handshakes and lobbying power in the world can't stop the tide of public opinion. They can't stop shame.

We're social creatures at the end of the day. And at some point, a marginally higher tax rate is easier to live with than constantly being shamed everywhere you go. The point that I was making is don't underestimate it. And the reason I say it's funny isn't because the point's funny, but because the timing is funny.

Little did I know that while Peter and I were talking about this in theory, on the other side of the world, a bunch of university graduates made it work. 

Last December, Time Magazine selected its Person of the Year for 2025. And it was this time, it was the architects of artificial intelligence.

Interesting

If AI in its current form was truly inevitable, if public buy-in didn't matter, and if people's repulsion to all this greed meant nothing, there'd be no need for Eric Schmidt to tell people to get in the rocket ship. The reality is none of this is inevitable. And this brings me back to the first thing that Peter said in our chat, that his brain is very literal 

And when I look at evidence and the evidence is overwhelming, I have a really, really hard time looking away from that and like pretending that I don't see it.

He can't ignore it. He can't look away. It is evidence. Well, I think that's why that clip is going viral because it is evidence of the big galvanizing pro-humanist effect that AI repulsion is having on people. I still don't think it's AI neural networks specifically, but I think it is the fact that they empower and accelerate greed to such an exaggerated degree that people are just done.

In the tragic words of Howard Beale, "They're mad as hell and they're not gonna take it anymore." And in the less tragic words of my friend Peter 

There's an alien invasion that's happening and we just happen to be making the aliens. And I think when people really see that, I think they really see what is happening and what these companies are doing, they'll go, "Oh, yeah.

No. No, no. No, you're not doing that. You're not doing that to me. You're not doing that to my family. You're not doing that to my kids. You're not doing this to like everything that I love." There are some people in Silicon Valley that are like not all that stoked to be human and I really don't like that. I think that's really, really bad.

I don't know. Being human is neat. Being human is really nice. I want to continue being human. Yeah. I would like our, our messy and complicated and, flawed, deeply flawed little species to continue. I'll take, I'll take our flaws over some mechanical perfection any day

Exactly the same technology can be used for good and for evil. It is as if there were a God who said to us, "I set before you two ways. You can use your technology to destroy yourselves or to carry you to the planets and the stars. It's up to you

the things that are considered AI are not always considered AI. So for example, the title of my, forthcoming book is Deep Unlearning, and this is a play on the term deep learning, which is a technique which is currently basically synonymous with AI. But back in the day, deep learning was not considered AI, so much so that, the researchers in that specialty were having, their academic papers, rejected by the prestigious AI, conferences.

And so actually, deep learning was a rebranding from another term, artificial neural networks, to avoid that stigma. So not only do we have this marketing term that, is a hodgepodge of disparate techniques or tools, and in fact, it's not even sometimes it's techniques, sometimes it's tools, sometimes it's products.

but, but also not all of these things are always called AI, right? When one of these subspecialties is really hyped up or has some better performance on a specific benchmark, it gets elevated to AI, and then it gets downgraded not to AI. And there are other such examples, like expert systems of the '80s.

That was what was synonymous with AI in the '80s, and now nobody calls expert systems AI. So, what happens here is that it makes it so difficult to have, conversations that are grounded and specific and, have specific conversations about what are the harms, what are the benefits, what approaches are good, what approaches are bad?

Because when I talk about some things that I'm really against, ChatGPT things, then people are like, "What about a particular system that is used for early diagnosis of cancer patients?" And, I have to say, "Well, that's a completely different system. It doesn't even have to be built in the same way that these big, models are being built."

Timnit Gebru, can you talk about f- co-founding Black in AI? but start off by talking about why you were fired from Google in 2020, the issues that you were raising there and the position you had there. 

so yeah, I co-founded Black in AI way before I even joined Google. my current institute is called the Distributed AI Research Institute, which I founded after I was fired from Google.

but, I, I started Black in AI in, around 2016, 2017 when I, attended, I started seeing, simultaneously the lack of Black people in AI. And so I would go to academic conferences, and you'd have about 5,000, 6,000 people, in those conferences intern- and only a couple of one or two, a handful of Black people.

And at the same time, we had, some of these, systems, some of the kinds of systems that Heidi w- was talking about. For instance, there was a ProPublica article from 2016 talking about pr- a company purporting to determine someone's likelihood of committing a crime again. And so th- this company was saying that they built a software that can, tell whether a, a person released from prison was likely to commit a crime again.

And the ProPublica article was talking about how, These systems were more likely, even, to label Black people as, criminals, potential criminals than, than white people. And it was very scary because judges were already using the outputs of these models to, in their b- decisions about bail, decisions about, how long someone should be in prison for.

So the dichotomy of seeing the lack of Black people in the field and also the kinds of claims that were being made and the kinds of things these tools were being built for was very scary. And so that's how I, I decided to found Black in AI. So by the time I started working at Google in 2018, I was a very known quantity.

I had already, bl- founded and led Black in AI. I was working on uncovering all sorts of issues in the field. my collaborator, Joy Buolamwini, and I had written a very, a paper that was, pretty-- it made the rounds, and was highlighted, and all, all over the place and even changed policy, showing for the first time that automated facial analysis tools like face recognition, for instance, tools were, had much higher error rates for darker-skinned women than lighter-skinned men.

And, the darker and darker the skin, and especially for women, the higher and higher the error rate. So think about these tools being used, on CCTV cameras to identify so-called criminals, et cetera. and the people who are dark- of darker skin would be more likely to be wrongfully identified, in this case.

And so I had already worked on this these kinds, uncovering these kinds of issues. And so I was hired at Google to co-lead a team, called the Ethical AI Research Team, which was a small research team with, Meg Mitchell, who was also later fired. And, the-- there were many issues. it was one issue after another.

This was in the middle of the Google walkout, where twenty thousand women walked out because, we found out that Google had paid Andy Rubin ninety-one million dollars, after allegations of sexual misconduct. and so it was a co- a, a combination of a whole bunch of issues, and it was at the height of Black Lives Matter, the Black Lives Matter movement in 2020 And at this time, a company called OpenAI, that was, getting very famous, but not as famous as it is today, came out with a large language model called GPT-3, which later became the backbone of ChatGPT.

But, ChatGPT wasn't released yet. And, GPT-3, was really hyped up in the mainstream media, and, OpenAI was making all sorts of claims about how powerful GPT-3 was. In fact, actually, they had said that its predecessor, GPT-2, was too dangerous to release because it's so powerful. and so, the GPT-3 was a large language model, and we can just define what large language models are.

they are models that are trained on vast amounts of textual data on the internet, and they're trained to, if-- calculate the most likely sequences of, text based on the training data. So if you use large language models to generate text, you are, trying to, generate the most likely sequences of, text given your training data.

the fact that, OpenAI got so much airtime for GPT-3 meant that all of these companies wanted to build similar, models, and they wanted to build larger and larger models, which means that, they were guzzling all of the data on the internet. and they were, consuming, using huge amounts of computational power even then.

and so we were very worried about this race that was started. Every company wanted to have the largest model, and we were asking, "Why do we need the largest of anything? What problem are you trying to solve?" And we warned about-- We wrote a paper, my collaborators and I, warning about the dangers of large language models, and we said, "How big can large-- can language models be too big?"

So the first, issue that we mentioned, the very first one, was the environmental and financial cost. And you can see how, six, seven-- six years later now, that is not only true, but even worse than we said in the paper. We talked about how, the carbon footprint of these huge datas, the, the, the necessary computational power, necessitates huge carbon footprint.

And even if you claim to have data centers with renewable energy, that, energy is going away from heating people's homes, for example, and towards training these models. and the financial cost that shuts out, anybody who doesn't have the wealth to, participate in this work. We also warned about hege- perpetuating hegemonic views.

the claim was that because these models are large, me- which means that they are using huge datasets, then oh, they have all of human knowledge, in these, in these datasets, but that's not true. the, the internet pr- represents hegemonic views. It does not represent views of everybody in the world.

A lot of people are not even on the internet. We even know articles like Wikipedia are heavily biased. It's a he- overwhelmingly mestern- Western and male. and, we, we also warned about... And this is very related to what Heidi was talking about earlier. We talked about how because large language models, are, tend to output fluent and coherent text, this can be very deceiving.

The majority of public AI discourse tends to revolve around a handful of AI companies: OpenAI, Anthropic, Google, Microsoft, NVIDIA, Groq.

It doesn't tend to revolve around AI safety scientists So Peter's job might not be immediately familiar to everybody, but one way to think of what Peter does is he is to these companies what a climate scientist is to big oil companies 

I find it really, really messed up that like dozens, maybe a couple of hundred people in Silicon Valley are making decisions about all of our futures.

And just like a climate scientist, the looming catastrophe that he studies and tries to steer us all away from is in many ways a global threat because of careless corporations, the extreme hubris of a handful of people, and plain old greed. And just as these things are an accelerant on AI as a problem, AI is an accelerant on every other problem.

Weapons, makes them worse. Ecological catastrophe, makes it worse. The mental health crisis makes it worse. So Peter's job is to get people asking questions like these ones. 

Can we slow down? Can we understand the science more? And can we like listen to the public? Like everyone hates this, right? It is 

one of those unpopular issues.

I, 

I think I hate it for a lot of similar reasons to the public and some like extra ones. Yeah. But I'm like an extra hater, And also I feel like really heartbroken because fundamentally the ideas are really cool. Like neural networks are really interesting and they're really cool and they can be applied in some like pretty phenomenal ways.

All right, so first things first, the term AI is broad. When I say AI, and when Peter says AI, when Eric Schmidt says AI, when you say AI, we could all be talking about different things. So in order for any of this to mean something or have any utility, addressing that broad term AI is the first thing we've gotta do.

When we talk about AI, we really shouldn't be just going like it's just three companies or four companies or whatever. AI as a thing has been a thing that we've been trying to do for thousands of years. People have been dreaming of building smarter than human machines. We have myths going back literally thousands of years.

When it comes to the actual tech, the idea of neural networks goes back to like 1942, 1943, right? Like the mathematical model of an artificial neuron goes back to the 1940s. AI has been a thing for a lot longer than Sam Altman's been alive. To equate all advances in AI with these like Silicon Valley companies does honestly everyone a disservice.

I really like the way Peter put this, that it does a disservice to everyone. But I think it's worth pointing out the very, very notable exception of who it doesn't do a disservice to, who benefits when these terms are vague. It's the Eric Schmidts of the world. And this was exactly what bothered me so much, that he saw this incredibly vague term and instead of defining it, he hid behind it.

Because if your label conceivably contains curing cancer, and if your actions are I build machines that kill people, why the hell would you ever define that? That is a terrible PR strategy. 

People are just, not convinced that this technology's legit. Yes. And I need to convince people, "Hey, this technology is legit" 

Yes

it can be really good for the world. 

Yes. 

And because technology is omni-use, it can actually be really bad for the world. Yes. And how the techno-feudalists in Silicon Valley are currently building this technology, it is more likely to be bad for the world than good for the world. 

On the positive side, even those basic selling points of- I can see a world where we have an incredible productivity boom

aren't a guarantee that the world will get better. 

In a decent society, just increasing productivity ought to be a good thing. But in our society, it's not necessarily a good thing to increase productivity because it might make the rich richer and the poor poorer. 

I interviewed Hinton for this video that I made, and I needed the, the clip, but I also just wanted to check if I was crazy.

Because I was like, "Hey, just, like, how seriously do you take this, the risks from AI?" 

There is something funny about wanting to check that you're crazy so you call up the guy who invented the technology. 

Right. it, it is definitely a flex to be like, "Hello, Nobel Prize winner. thanks for taking the time."

But yeah, I was like, "Hey, how seriously do you take the risks from AI?" And he went, "Oh, really seriously. It's either climate change or AI, and I think AI's going to win." 

I think right now it's probably a race between climate change and AI, and AI's gonna win 

Kinda wanna clarify that Peter isn't actually a doomer.

He is passionate about science, about teaching, about hope, and about the welfare of other humans. And while we're on the topic, it's worth investigating the term. AI doomer might sound anti-AI, but the doomer narrative ends up helping the AI companies by treating AI as apocalyptic and inevitable. The power to doom the world becomes a sales pitch.

Are you trying to learn computers? Are you finding it frustrating? We have a simple method. We just tell you what keys to push. 

AI doom disempowers the general public to engage in any protest or resistance. 

There are two ways in which people are controlled. First of all, frighten people, and secondly, demoralize them.

In reality, Quit GPT has really hurt open AI. Palantir, with the help of public sentiment, is losing contracts left, right, and center, and local communities have delayed, and in cases blocked, $64 billion of data center projects. But where this story gets weird, and honestly a little frustrating, is that just because the AI is powerful narrative has been weaponized, this doesn't automatically mean that AI isn't powerful.

And so you end up with this situation where the Silicon Valley CEOs and the scientists who study the stuff, who really don't like these CEOs, keep telling us the same thing. 

AI is incredibly powerful, and also how we're currently implementing AI is really dumb, and it's really dangerous. 

I realized pretty quickly into talking to Peter that there was a problem with my question, AI, what could go right?

You couldn't answer it without also implying what could go wrong. 

AI is getting better and better and better at biological stuff, and that's really good when you're doing drug development. That's really good when you're trying to find proteins that are, like, malfunctioning in a certain disease or whatever.

It's kinda bad if you're trying to make toxins. 

It is. 

Or if you're trying to make novel viruses. 

Yeah. 

AI aided drug discovery is already a massive thing, and it's undoubtedly very, very effective. 

Mustafa Suleyman, his big argument, what he saw as the biggest fear was rogue people having both CRISPR and AI and making bio weapons in their garage.

That feels right. 

Yeah. 

That feels right. That feels like we're approaching that. 

Come on, guys, don't do that. But again, AI rhetoric, we're not there yet. Well, we are with toxins, but I feel like the word yet is doing a little bit of heavy lifting. AI isn't magic powers. There's a lot of steps between you and a bio weapon.

So what about something closer to home? How should we think about the problems that are already here? There's all the cognitive stuff, psychosis, the effects of persuasion, computational propaganda, data centers, the environmental impact, IP theft, surveillance and military issues, the potential job apocalypse which manifests as what jobs will it take or make less stable, and what people can exploit this jobs are scarce narrative 'cause making workers feel grateful to at least have a job is the conditions for somebody to screw over those workers.

There's the fear of the bubble bursting.

What happens there? Scams, deepfakes, social engineering, a ragtag team of misfits breaking into a foreign government's computer system. I don't know. 

The cybersecurity stuff is terrifying. Our world really does run on computers. Our energy grid, our banking system, pick a thing and it's connected to some electronic thing that manages it.

We're now at the point where, one person One person using a commercially available tool was able to hack into the Mexican government and steal like 150 million taxpayer records and a bunch of like really sensitive, really private information. 

there are tons of studies about how AI's use in a classroom setting dulls critical thinking, reduces the amount of non-verbal intimacy, which makes up a majority of effective teaching, things like tone, cadence, eye contact, et cetera, which then leads to student alienation and even still discrimination because AI has to be racist, too.

Stop telling people from marginalized communities not to use AI. For many of us, it is the first time in our lives that we're able to use a free resource to explain things to us in a simple way, without judgment, and most importantly, without the price tag. This is the first time that there's been an opportunity to level the playing field even a little bit using technology, so let people use the tools.

For many of us, it was the first time we were able to have something explained to us in a non-judgmental way. Have you never heard of Khan Academy? Have you never heard of Crash Course? Or better yet, a book. Books are fucking free, bro. Listen, walk into a library, walk up to the person at the counter, ask for a library card, give them your name and email.

They will give you a library card for free, and then turn around. Behind you are books. Did someone pay you to make this video? 'Cause sometimes I feel like people are just saying shit, and it's like a prank on me personally. Wait, I'm sorry. Are you under the impression that AI makes you smarter? She said the rich and wealthy are already using AI to, optimize their businesses and shit.

Who do you think that hurts? Who do you think that's replacing? 

Sure, a lot of people don't have access to libraries or a network of peers or teachers who give a shit, particularly marginalized folks or people in rural areas. But like he says in that video, YouTube exists where you can find resources of altruistic, qualified people doing what ChatGPT claims it can do for free in a much more entertaining and engaging way in under 10 minutes sometimes.

Besides, countless channels have college-level summaries and analyses written by experts that was probably used to train AI models- sans the hallucinations. With just a little digging, you can also find academic papers other people have written that analyze most of the topics and subject matter you're gonna have to learn in school.

Just type .pdf after after a search. Something that people often overlook when making this argument is reckoning with what people did before generative AI. It's like the pro data center argument. We need them for cell phones and internet too. We were doing just fine with the amount of data centers we had just a few years ago.

We don't want or need more for a technology that has yet to prove its value. And maybe making that comparison when it comes to education is a little unfair, because this country hates people trying to better themselves, and has always made education extremely inaccessible for way too many people. But the answer isn't to outsource it to a flawed machine that incentivizes laziness.

Companies and governments could take that data center money and use it to fund rural areas and provide resources to those marginalized people in the form of human teachers that actually know what the hell they're talking about, and care about their students, who will look them in the eye and tell them they're doing a good job.

About a week after I started working on this video, Dolly Parton died. She was someone who I can confidently say everyone loved and respected across political and religious lines, who all pretty much agreed that her philanthropy was as important a part of her character as her Southern rhinestone charm was.

For the past three decades, her Imagination Library has given out over 300 million books to children from all different socioeconomic backgrounds, just because she wanted to help and understood the value in providing resources to developing children, and not a shortcut. It was also funded in part by private and public donations.

And what some states have done over the last few years? Defunded it. Defunded Dolly Parton. I wanna defund a lot of things, but I don't wanna defund Dolly Parton. And I can't prove it, but I wouldn't be surprised if some of that money found its way into the pockets of OpenAI and Oracle. On the front page of their website, there's a quote: "I know there are children in your community with their own dreams.

They dream of becoming a doctor, or an inventor, or a minister. The seeds of these dreams are often found in books, and the seeds you help plant in your community can grow across the world." When she says books, I think she also means an education, the ability to comprehend the complexities of the world around you, to sort between fact and fiction, and truth and lies.

And learning to read early is a first step towards that comprehension. Without her support of those future doctors and inventors, and ministers, I guess- By very simple means, our world would be a worse place, and if one person can use all that fame and money to do good, who isn't even a billionaire mind you, then countless others with just as much influence and wealth can as well.

But they choose not to. Governments and trillion-dollar, trillion-dollar companies can invest in communities, not by offering a snake oil shortcut that has no proof of effective concept. Handing off teaching to an AI model is short-term gains in exchange for long-term disadvantage, i.e., being able to figure out when you're being taken for a ride.

People with low literacy skills tend to have low critical thinking skills and an inability to process complex linguistics, and thus social contexts that manifest in frustration, anger, and anxiety, ultimately alienating them even more. Basically, using AI as an educational tool is doing to Gen Z and Alpha what lead poisoning did to Boomers, and what microplastics are probably doing to Millennials.

Whatever's going on with Gen X is their own fault, though. Maybe it's meth.

if you've been reading local news, you might know that water treatment plants have been being cyber attacked at the highest rate ever. And that's very little to do with AI. It's to do with the Iran war. But the way in which they're being hacked is a lot of these systems are sat on the open internet with default passwords, passwords that anyone can guess.

Someone can literally go and log into the infrastructure that controls a town's water supply, the water they need to survive, and take control of it That's not an AI issue. That's an infrastructure security issue. That's an ICS issue. That's a cybersecurity issue. It's a resiliency and a disaster recovery issue.

It's pretty much everything but an AI issue. And we're trying to focus on problems like that, problems like defending networks, defending critical infrastructure, and the AI bros are coming out like, "Look at me. I've just come up with this new hypothesis for how AI wipes out all of humanity." And for some reason, it involves cybersecurity, which is mind-boggling to me because while cybersecurity is an issue that is near and dear to my heart, it is not the worst thing.

Like cyber attacks are not the worst thing that can happen. In fact, Anthropic came out just last week with a report that terrorist organizations were using their AI model to vibe code ballistic missile guidance systems, a phrase that I could have never in a billion years imagined myself saying. And of course, one single missile could kill more people than every cyber attack in history combined.

But for some reason, they always just come back to focusing on cybersecurity. And I think it's because they're just so out of touch that they really cannot imagine harm beyond some computer systems getting hacked. They cannot comprehend the idea that there is more to the safety of the human race than stopping computers from being hacked.

And it's actually something that I've noticed myself in my own research. Anthropic is very notorious for having very strict guardrails with regards to cybersecurity, to the point where I was analyzing an attack, like an attack in a client network, and I asked Claude, "Hey, what does this code do?" And it was like, "Oh, oh, sorry, I can't tell you that.

if I tell you that, it's, it's hacking. You'll, you'll understand hacking, and then you can hack." And I'm like, "You cannot be serious. Like you're refusing to help me analyze a security breach because you think telling me about how it happened would help me hack. Like I can't go online and Google how to hack."

And bearing in mind, I'm in Anthropic CVP, which is their cyber verification program, which gives, verified cybersecurity professionals access to less restricted versions of their model, and I can't get it to do defensive security work. But then I was like curious. I was like, "Well, is this the same for every area?

Like what if I was asking about how to build weapons?" So I went and I started asking it questions about how to build weapons. And like to my absolute shock, it was just happily going along. It was like, "Yeah, you can do this. You can build this." And it was telling me how to build weapons that can end a human life.

But analyzing a cyber attack was just way too dangerous. And it brought me to the realization that Anthropic does not understand risk. They do not in the slightest understand risk. They have, for some reason, this insane fixation on cybersecurity and cyber risk at the cost of everything else. if I saw terrorist organizations using my product to build missiles, that would be my next blog post.

It wouldn't be, "Oh my God, what about a big botnet, like a, a botnet that is so big it takes over the entire internet?" I'd be like, "Yeah, the, the missile, dude. the, the people are building missiles." But the fact that terrorists being able to make advanced weaponry is being taken less seriously than completely made-up scenarios in which, a lot of computers get hacked, it, it's mind-blowing to me.

I just... I cannot wrap my head around it. These people are so deeply unserious. And of course, when you ask them, "What is, what is your trajectory? Like, how do we go from where we are now to this, insane hypothesis where AI wipes out humanity?" Their answer is usually one of two things, AGI or RSI.

RSI is the idea that an AI model could build a better AI model, and then that AI model in turn could build a better AI model, and then suddenly the capabilities just go parabolic, and suddenly you have a super intelligence that's outside your control AGI does not exist. No matter how much the marketing departments try and claim that they have achieved AGI, they have not.

And the same with RSI. They have not proved that meaningful recursive self-improvement is possible. So essentially, it is a cult. We have built these ideas of these AIs becoming gods, unstoppable superintelligences, all-knowing, all-being. They can hack any system. They can destroy anything. And we're fighting that instead of the very real risks that we actually have.

ChatGPT is giving people literal psychosis. We have terrorists making missile guidance systems. We have people using AI to aid in phishing, aid in writing malware, like actual real threats to deal with. I am deadly serious when I say this is a cult. It has all of the makings of a cult. But instead of an omniscient god that has existed since the beginning of time, it's humans.

They made the god, and they lose control of the god. And we're treating this as if it is a real safety issue that, needs real attention and real experts dedicated to instead of actual problems. some guy's doomer god fantasy is what we need to be spending our time on. I'm just gonna be honest.

I'm sick of it. I am sick of listening to these people. I'm sick of the Dario Amadeis. I'm sick of the Sam Altmans. I'm sick of these random researchers coming out of the lab like, "Oh my god, guys, I've been working on the orphan-crushing machine, and you will not believe this. It crushes orphans. I need to tell you about how bad it is."

And it's like, yeah, dude, we've been saying this the whole time. We were all like, "AI sucks. Please stop shoving it down our throats." And also, you're being insanely reckless with it. OpenAI went and they made the most terrible sandbox in history, and then it, their AI agent got out, and it hacked a site, and they were like, "See?

This is how dangerous AI is." It's like, no, this is how dangerous not sandboxing technologies is. This is like the CDC walking down the street with a comically large vial of smallpox, dropping it in a crowded area, and then being like, "See? This is why we need to care about this god we're making that makes worse smallpox."

It's entirely insane. I 100% agree that we definitely need to be focusing on how absolutely reckless these AI labs are, but not because they're making gods, not because they're building superintelligences or AGI, but because they're making systems that can cause harm, and they can build weapons, and they can give people mental issues, and they can hack, and they are controlling them in the most reckless way possible.

when we talk about superintelligence exceeding humans at all cognitive tasks, the question is what do we mean by that to a degree? Because it's, this is about us in the real world being, with the physical environment, we're talking, we're communicating. And, and we- I was with my, with my little nephew who's 16 months old, and she said, it's 3:00 in the morning, an unfamiliar bo- baby's crying.

Work out what's wrong and safely get them back to sleep." And these are the... she said, "These are the sorts of questions that would be asked. Is that a cognitive task?" Do you see what What, what do we mean by superintelligence exceeding humans at all cognitive tasks? Bearing in mind that completely, it sounds like it...

some people are like, "That's left field," but it's about the interactions in the real world that define us. 

Totally. Yeah, I would say, if, if you're asking my definition of superintelligence, I would say the AIs would need to be able to succeed at that before I would call them superintelligence, at least as we defined it in the book.

and so it would need to be the case that, if you put this AI in a robot body, it is able to go, figure out what's wrong and, and, get your nephew back to sleep. and I think that if an AI is smart enough, it can figure out how to run that robot body, figure out what's going on in a human, figure how to steer it, figure out how to, how do like diagnose, whether the child needs burping or whether they just need to be rocked back to sleep.

the, the flip side of this, though, is that, If we define super intelligence as better than humans at every cognitive task, then that means we don't get to be surprised by the world being changed by things that don't technically count as a super intelligence. what, what might matter here is not whether, when the AIs are better than the best human at everything.

What might matter is when are the AIs able to do AI research at a superhuman level and self-improve, When are the AIs able to do their own novel scientific and technological development? When are the AIs able to run a robot factory to build more robots, and then use those robots to run laboratories and start inventing their own tech?

That could happen before AIs are better than us at everything and, and that's the line that determines our fate. 

And just a couple of final things. what, one, is, is... Would it end up being exponential in theory, the improvements that would happen, and with no limit, and therefore the gap between those computers and us would be like we are ants essentially to them?

We, we... And there's no way they can outmaneuver us. Or we can outmaneuver them, sorry. 

Yeah, it's not without limit. There's definitely limits. the light speed limitation fundamentally says you, you can't get that much machine in one spot con- communicating with all parts of the machine fast enough.

And there's, there's fundamentally limits. the issue is that the limits are super high, if, if, If someone's like, "Hey, we're testing a nuke in this village, maybe you should leave." And you're like, "Oh, well, a, a nuclear explosion can't run away forever. There's limits on how fast and how far, a chain reaction can go.

Eventually it's gonna run out of fuel." yeah, also please evacuate the town. We're, like, dropping a nuke here, the, the fact that there exists limits does not mean the limits are low. And with intelligence, it looks like the limits are probably really, really high. humans just have a lot of limitations in our intelligence.

We're running on neurons which are much slower than transistors. we're running on brains which are small compared to how big a data center can be. we have pretty low, access to power i- in terms of, like a, a human runs on about as much electricity as a light bulb. and whereas, whereas these AIs are running on, as much electricity as a city.

They're, like, radically less efficient than us. We know this. But, what's the limits? The limits are huge. The limits are, like, AIs that think a million times faster and can make a ton of copies of themselves. and, it's, the, the, the, the issue with the runaway AIs is not that they run away forever.

It's not that, they become an all-powerful god. the issue is that they could vastly outstrip humanity. And if they don't care about us, and they're the ones running all the technology, running the show, then we're likely to die as a side effect. Again, not 'cause they hate us, but just because, they build their own technology, their own infrastructure, their own automated data centers, they make more data centers and, and, they're like, "We actually wanna run the f- the, the planet really hot 'cause that's the most efficient for running data centers."

And we're like, "We are going to die if you do that." And if they don't care, you have a problem. And, we don't, we don't know how to make the AIs care, which is in some sense the crux of the issue. 

Just very lastly, what, what if the real risk is actually by k- acting on the mounting concerns, billions of people die in that superintelligence could cure cancer, abolish aging, basically d- deal with all the issues that are actually killing humans, and make...

we could end up, if they abolish aging, people could essentially immortality. What about that? What, what if that's the real risk, that we're gonna stop development which, which saves us from all these terrible fates? 

Yeah, a lot of people say, "Oh, we have to either stop and never get this advanced technology or gamble right now with these, AIs nobody understands and we clearly can't keep control of."

And this is just a false dichotomy. there's a third option. You don't need to choose between gambling humanity at, 10% or 90% odds, whatever you think it is, and never finding some way to get this wonderful tech. We can just stop the crazy AI race and, find a way to get that tech more safely.

Frankly, you can continue a lot of the cancer research without, trying to make a superintelligence. we can just deploy current AI tech towards drug discovery and cancer research without needing to make a smarter swarm. And if you're like, "Well, we need the smarter swarm to cure aging because, that's the only way to get it smart enough to cure aging," well, it's not gonna cure aging if you make a smarter swarm and lose control of it and it kills us all, right?

sometimes I feel like we are in this bus racing towards a cliff, and I'm like, "We need to stop the bus before you go over the cliff." And people are like, "Well, there's a huge pile of gold at the bottom of the cliff, though." And I'm like, "Yes, I agree that there's a huge pile of gold at the bottom of the cliff, but slamming into it at terminal velocity is not a great way to add all that value to our economy."

I'm not saying we can never get it. I'm saying we gotta stop the bus and find some other way down. 

Now, Section C, The Grift, The Bubble, and The Doom Marketing

It's easier for us to imagine Terminators than to wrestle with the idea that these companies are gonna blow up the economy, right? It's like there's, there's a weird- Yeah ... there's a weird like self-soothing like, "Oh, I'm worried about the Terminator," this like mythic story, rather than the old boring story of this isn't the Terminator, this is the big short, right?

Yes. I think underneath all of this is that like the actual thing, the doomsday situation here might actually be that these companies end up cratering the entire economy. For those who, who have been focused on the Terminator movie version of AI and not on the big short version, of the AI story, who may think that the big short version of the AI story has nothing to do with them 'cause they, all they use is ChatGPT, just remind us of like that happening concurrently here.

Like how the actual doomsday story for all of us, for everyone listening to this might be an economic story. 

So from the very beginning, large language models have been sold on what they might do rather than what they do do. They are useful for search, but again, they hallucinate mathematically certain to make mistakes.

Useful for coding. How? The more complex the coding, the less reliable they are. Great for very specific functions. The more you expand, the more unreliable and more unstable they become. They're also incredibly expensive both to run and to train, costing billions of dollars a month at this point, I think for Anthropic and OpenAI.

And as a result, they require over a trillion dollars of capital expenditures from Microsoft, Google, Amazon, Oracle, CoreWeave, and a bunch of other companies to the data centers. The reason these companies want you to think about superintelligence is so that you don't think about the rest of it, which is this is cloud software.

My girlfriend actually made a point about this, which is if the... and this is a problem in the media, in investors, in everyone thinking about AI. If this isn't the Terminator, if this isn't the path to superintelligence, everyone has to think, "Oh God, we just spent three or four years going insane about cloud software, spending a trillion dollars, making our stock market functionally dependent on the earnings of tech companies for cloud software that's really unprofitable, unsustainable, doesn't really have any effect on productivity data, like doesn't actually appear to be helping people that much.

The data says that actually they make some engineers slower." If you can't see it as this big scary thing- which is a way of rationalizing all the expenditures. It's very, very dark and sad, and tells you a lot about the world. It tells you that the media does not care about getting it right.

Fell for it again award for everyone. That they... And the thing that Matt was talking about was very, very accurate, where it's they hear these people and they go, "Oh, they're rich. Oh, they're working at the companies. They must know." It's weird. They're like, "Well, these people who have an obvious financial incentive and a obvious peer-based incentive to say these things, they'd never lie, they'd never overstate."

And the problem is, when you don't really interrogate the underlying technology or the numbers, all you do is just summarize. And because AI exists as this very spec- good at very specific things, but in the, in the round is questionably useful on, a general purpose level Suddenly you just end up publishing whatever they want.

They're just like, "Well, it's got... We've had this much improvement, so it will improve this much in the future. Invest now." It's much easier to believe that all of this data center capacity was built because all of these smart people have an idea and they're building towards it than it is to accept that we're at the end of tech's hypergrowth era.

The tech industry has run out of big ideas. They don't have a new Google search. They don't have a new Facebook. They haven't had for years. This is their last shot. And most of the revenue for AI, by the way, comes from Anthropic and OpenAI. So again, people say, "Well, they wouldn't invest all this money for no reason, would they?"

Well, 

it's not for no reason. They were just wrong. But again, s- like Stoller said, it's easier to think, "Well, they're all building dangerous superintelligence," than thinking, "Oh, everyone was wrong. The media got it wrong. The investors got it wrong. Everyone got it wrong." And now while this might not be great financial crisis bad, it really comes down to how private credit goes, because we don't know how big it is due to the nature of private.

This is a worse than dotcom scenario for the markets. On top of the fact that afterwards, big tech does not have a next big thing. 

Let's touch briefly on this idea that AIs are gonna be replacing humans for work because, of course, I think for a lot of people, this is the big fear that they're, living with in their day to day, is this looming idea that there is an AI coming for their job. I'm thinking particularly when it comes to younger people.

There was a McKinsey report saying, I, I think like 51% of firms are already, cutting their new hires of graduates. so, so what do you say to people who are watching this who are saying, it's all well and good saying the AI scare is overblown, but we're already seeing the job market change, and I'm already seeing, for young graduates, that it's really hard to even get a job now because those lower level entry jobs have been- Yeah

replaced. 

So in the, in the history of automation, this isn't the first time we've had one of these struggles between, labor and capital over automation. And where automation is driven by labor, it is typically adopted in service to making a better quality output. basically, workers pick up a new tool if they think it'll let them do a better job Which you've all done that, right?

And something that we do, I don't know, if you paint Warhammer miniatures, you don't go out-- No one makes you buy a new brush, but you might be like, "Oh, this brush is gonna let me, make the world's tiniest eyeball dots." and when capital drives automation, it is typically in service to increasing throughput, making more, right, per worker.

And where capital has market power, where they can force consumers to accept an inferior output, so think here of like AI chatbots replacing customer service, right? so customer service sucked before, and now it's the, the bad joke, right? And, and, particularly like you can tell that they didn't wanna do good customer service 'cause they went to the Pacific Rim, and they hired like smart people, like people in the Philippines and India who had gone to university and worked really hard, and then they said, "You are not allowed to solve anyone's problems.

You must follow this script, and the script does not have any place that it ends in which someone's problem is solved. Your job is to speak reasonably good English and get shouted at." And of course, a chatbot can speak reasonably good, English and get shouted at more, far more cheaply than anyone in the Philippines or India, so we're moving that to them.

And so if you have market power, you can force people to accept inferior output. I'm, I'm trying to get someone at the Royal Mail to f- actually deliver a parcel for me, and all they do is put red cards through my mailbox, and it's impossible to get anyone on the phone, right? And when I do, that person's job is to just get shouted at me because they manifestly cannot solve my problem.

Thank you for privatizing the Royal Mail. This is great. I'm so glad we're exposed to the efficiencies of the free market. So, that means that we are now living in a moment in which senior workers are w- being asked to do more, right, with fewer colleagues. they're being asked to effectively move from doing the part of the job that requires discernment, judgment, creativity that is satisfying and just becoming someone who like babysits the computer, who marks the computer's homework.

and because they can increase throughput just by having one or two senior workers who would've historically been, at least in line for some market power because they're senior, which means that there's demand for their services. But now if you can say, "Actually, we're just gonna make everything a bit shit, and so we don't need as many workers 'cause we don't need anyone to do the quality assurance anymore.

and we're gonna, therefore have far more labor competition among senior workers." And actually it's quite nice that there's all these junior workers who are unemployed because they become a standing army. You wouldn't wanna be-- You wouldn't wanna have to join those junior workers out on the breadline, so you better accept the lower wages or worse working conditions or longer hours.

Remember, working longer hours for the same pay is taking a pay cut, right? It's your, your hourly rate has gone down. And so this is a moment in which capital has found an opportunity to dis- to discipline labor. And because of the AI bubble, it's also a moment in which firms can announce that they're making swingeing headcount cuts without the market making the inference that they're doing worse.

Normally, if you make thousands of workers redundant, it's because your sales have dropped off. but now you can say, "No, it's because the AI has solved our problems." And the countervailing trend with AI is that lots and lots of workers are being hired back on Because it turns out that they're cutting past the bone.

Hmm. 

And I'm quite worried about this because just rehiring those workers doesn't mean that everything goes back to normal. There's a really important part of industrial process called process knowledge. So we, we're all familiar with IP. IP, you could broadly say, is anything you can write down about how a worker does their job, hang up in a filing cabinet, and sell with the company.

It's an asset. Process knowledge, sometimes called, tacit knowledge, although that's usually like what one worker has. Process knowledge is more all the workers in a, in a firm or across firms in a sector. It's the-- It's the little bits and pieces of lore that you could never write down.

Even if someone offered you $20 million, you couldn't write it all down. You certainly can't train an AI on it 'cause it's not written down. and without it, the company grinds to a halt. there's the machine, and the machine takes an input, and if you put the input in too fast, the machine jams. So first of all, there's some process knowledge about how fast to put it in, but there's also, oh yeah, the, there's the one person who knows how to unjam it, but if they're on vacation, the person that used to have that job retired last year and I have their mobile number, they'll do it for 50 quid, and that's the difference between whether the factory's running today or it's not.

To give you a very concrete example of this, a couple of years ago, there was a chip foundry making microchips. Chips came off the line. They were defective. They could only run at about half their nominal clock speed. So the person in charge of this had process knowledge. They knew that they had a customer, and within that customer, there was a faction, and that faction had been saying, "We should really make a low-end laptop.

There's a-- There's demand for this." And no one believed them, and it was too expensive to try. So they called up this s- this company, this, this customer, and they said, "We got a load of really cheap, slower chips. Do you wanna take them off our hands?" They said, "Yeah, this is what we need to convince the company to go out with a cheaper chip."

that product was called the MacBook Neo. It is the most successful product launch Apple has had in 10 years. They are now manufacturing low-end chips because they s- they sold out of the chips that came off the line that wouldn't run at full speed because it was so successful. That's process knowledge, and when you fire everyone, you vaporize the process knowledge because the way process knowledge carries on is through something like apprenticeship, right?

You go to work and people show you the ropes. They answer the questions. You learn that when this thing is stuck, you hit it like that. Hmm. Right? If you fire everyone, all that lore disappears, and then it has to be rediscovered, and that's the work of a generation. To do. And so that is going to make the firms involved significantly less productive, which means that our economy is gonna be slower and weaker.

It means the workers aren't going to produce the things that we need and love. It means the workers themselves will have worse jobs. It means there, there will be fewer profits for workers to lay claim on if they can find a way, for example, to unionize and make a claim on those profits, right? It's, it is a cataclysmic thing that is happening because of this.

You are right if you're a young worker to be angry about AI, but it's important that we distinguish between AI can do my job, which is a reason for people to invest in AI companies, and my boss can be convinced to fire me and replace me with an AI that can't do my job, which is a very different story.

We don't wanna help them raise money for the bubble because then the bubble goes on and gets worse, and when it detonates, it'll be an even bigger problem.

If I could, I would certainly slow down AI and robotics. 

Really? You want to slow it down because, you've been promising robot butlers since 2021. You said you'd have a functioning prototype by 2022 and a commercial release by 2023. You remember that Kim Kardashian Tesla bot commercial campaign? Yeah, those things were already a year overdue when that happened.

So far, all you've actually delivered is a demonstration of a remote controlled robot butler being operated by an actual human, which is, objectively worse than just a regular human butler. Oh, yeah, trust me, I would really love to deliver on, the promises I've made to my investors, but, I can't because, Actually, it's 'cause the technology is too cool and awesome.

Anybody else notice that whenever these AI companies release information or one of their employees leak information that their super powerful AI has escaped its sandbox and hacked some website, it always happens in a way that didn't actually cause any tangible or monetary damage? See, a conspiratorially-minded person might say it's almost as if they're providing us with the evidence that would call for regulation, but in a way that wouldn't call for any legal or fiscal accountability for the companies themselves.

Now, don't get me wrong, I would be super in favor of humanity deciding that we should definitely be pouring way fewer of our collective resources down the throat of the Slopinator 9000. But that's not what they're calling for, is it? They don't want a reduction of collective investment in the infrastructure of their technology.

They're just calling for a reduction of the expectations for an actual return on that investment. Now, when you look at the statement released by Anthropic's CEO demanding that we pace the frontier, one of the things he's calling for is an exception to antitrust laws. the laws that exist to prevent giant corporations from collaborating in order to monopolize their industries.

It also seems like the regulations he's suggesting would put the major AI corporations in charge of determining what the regulations should be, which would allow them to set regulations preventing competitors from ever catching up, that are too expensive for smaller companies to exist or compete in the field at all, and that would functionally remove open source AI models from the field entirely, completely securing total corporate ownership of the technology in perpetuity forever and ever and ever.

See, that same conspiratorially-minded person might say it looks like these companies are trying to get a get out of jail free card to not deliver on the amazing progress they've been promising without bursting the illusion that that progress is, in fact, just beyond the horizon. Now, selling empty promises is Elon Musk's bread and butter.

He's promised Hyperloop trains and underground systems to solve the LA traffic problem, and tourism in space, and colonies on Mars, and cities on the moon, and rocket planes that take you from London to New York in just a few minutes, and cutting trillions of dollars from the federal budget. The list goes on and on and on.

His entire fortune consists of people actually believing his empty promises. Tesla is profitable, but nowhere near enough to actually justify him being the richest man in the world, or even really in the top 100. And SpaceX, well, SpaceX hemorrhages money. This is not traditional capitalism, right?

Capitalists are supposed to make money by selling products to people for a profit. The product that Elon Musk- sells is not Teslas or Starlink, it's the stock in his companies. And the way he maintains the value of that product is by maintaining the illusion that the value he's promising is in fact just beyond the horizon.

Which interestingly works, and not just for him, right? If you've bought stock in Tesla, then you've made a profit from owning it. But there's no such thing as generating value out of thin air. It's pixie dust. It's fugazi. That value will be paid for by whoever is left holding the bag when the bubble finally pops, which I believe is what he and his fellow corporate AI compatriots are trying their very, very hardest to delay for as long as possible.

But if you are still a faithful believer, I feel like you should be reminded that as it currently stands, it looks like we're actually gonna get GTA VI before we get Elon Musk delivering on a promise

a thing I've thought, not an original thought, is we know that these companies know everything about us.

They can predict our consumer desires. They can look at our scrolling to know if we're expecting a child or if we're depressed or whatever. Surely they don't need the IDs of people to know that they're an adult, and surely they, they like, they know who you are. They know who the kid... they know all of this.

So it's not like they're doing this for safety. It seems pretty clear that it's a surveillance type thing. Which makes it a bit strange that the same skepticism applied to other areas of technological development is read as a primitivist conspiracy among the broader left by Lorenz. And this has been a thing I've seen Taylor talk about more recently, whi- which is the the left.

whenever people use the left, I wonder, is that a real, is that a real thing you're talking about? Is there a real thing called the left? Point to the left. in the room, I saw some of this interview that's being referenced, and a lot of this interview and things I've seen Taylor say recently refer to a backlash against tech by the left, and in particular, the online left.

I'll put my cards on the table. I'm, I'm very anti-big tech, a- and I'm not anti-big tech in a way that's even that unique to tech. I, I'm anti-big tech in the same way that I'm anti-big oil, let's say. Big oil, bad for the planet, like we're all gonna die 'cause of it, things of that nature. Big tech, I think is actively eroding our humanity in a way that few technologies have historically.

I think big oil is eventually going to erode the conditions of the biological possibility of human life, let's say, but the existence of big oil doesn't, in a really direct way, suck away at the fundamental marrow of what it means to be a human in the way that I think AI does and in the way in which I think Silicon Valley makes very clear they want to.

A-and it doesn't take a lot of time spent with the leaders of Silicon Valley, and by time spent with, I don't mean in their sex dungeons in San Francisco. I, watching interviews, reading their books, the, reading the things they write and the things they say, to get that these are people that don't give two fucks whether you live or die.

But, but back to this piece. In a recent interview with Max Tanney and Ben Smith on Semafor, Lorenz says, "So much of the backlash to AI is the backlash to Big Tech. It's people that are angry about the influence that tech has over their lives, and they feel powerless against those big tech conglomerates."

Yeah, most definitely. So much of the backlash to AI is the backlash to Big Tech. It's people that are angry about the influence that tech has over their lives, and they feel powerless against those big tech conglomerates. Lorenz is right in offering a diagnosis based on power. People are concerned about the asymmetry between communities they live in and firms coming into them, keen on advancing an undemocratically planned and privately managed build-out that's at times hysterically marketed as the prelude to the destruction of life as we know it.

This is important 'cause I think, and we talked about this last week, there is a hyperbolic reaction to things like the data center and the build-out of tech that's this person's gonna build the data center, and then the data center's gonna start the singularity or whatever, and then, and then the singularity is gonna, gonna kill us all with computer.

sure, I don't buy it, and if I'm wrong, I'm wrong. If I'm wrong, I'm wrong. But I don't think I'm wrong. But I think what it is for a lot more people, and this is what Edward's getting at in this piece, it's the feeling of disempowerment by people. When the, when the big data center gets built in your town, it's not the mo- for most people, it's not the feeling of, "Oh, no, they're gonna make the singularity that's gonna, that's gonna fuck my brain till I die."

It's that, "Oh, giant corporations run by people who live on the other side of the country get to determine what happens in my backyard, and I have no recourse or say in that. They're colluding with the local government who I voted on to represent me, and holy crap, it seems like political and economic interests are colluding together in a way that shows no care at all for my existence or life."

And yet, a few seconds after making that observation, Lorenz adds, "This is my big disagreement with the left. I think a lot of people on the left do not believe in a tech-forward future. They just wanna go back." Before we keep going, let's, let's all put our cards on the table, cards on the table day. 

As a person on the left, as a member of the online left, I am not someone who's anti-tech.

I'm anti-big tech or Silicon Valley. I'm anti-technological innovation being inherently tied to market forces and the military industrial complex in the same way I'm against, medical research being a part of a for-profit industry, The issue isn't the thing. It's the structural conditions that the thing lives in

And Finally, Section D, Fighting Back - Workers, Communities, and Alternatives

the tech bosses have, have left their carrot era and entered their stick era. And that is an explicit aim at disciplining their workforce that they view as unruly and in the way of their race to win AI. And, something that we, we try to chronicle in the book is pretty much every moment of tech worker resistance or collective action to push back on the boss results in retaliation from the boss.

And what we saw in, 2022, even up to today, is retaliation for a movement that spanned the industry. tech bosses were looking at Google, Amazon, New York Times, a- almost every tech company was doing union avoidance over the past few years. And they saw an opportunity that aligned with their fiscal responsibilities to discipline their workforce with mass- Hmm

layoffs, with return to office, with, p- basically precarity, which is their biggest tool to suppress organizing 

Yeah, and I, I, I think the, the, the way in which they go about this is, is also, is also quite interesting, right? And I think it relates to the, the rightward shift of the tech elites.

we-- if, if we were to go back to the early tech worker movement and take account of the kinds of things that tech workers were protesting against, they were like you were saying earlier, these social issues, social justice Me Too, LGBT rights, but as, as well as, anti-militarism, this kind of, this kind of thing.

And, And I, I, I think the way that the tech bosses basically interpreted this is that it was the kind of progressive values that, that they had cultivated within the tech industry that had caused their own employee base to turn against them. And so part of their retaliation against the tech worker movement is to, is to go against those progressive values, right?

Is to wage war on what they would call the woke, values, right? That, that had caused the mutiny in the first place. Right? And, and we can, we can kind of see that in, in terms of how, how they go about disciplining this workforce, right? when you think of, around 2022, 2023, that's that's a moment in which we start to see tech bosses really go all out on this attack.

And- Mm ... and part of this attack is attacking DEI, attacking sort of the AI ethics crowd, right? Attacking these havens where they believe these kind of progressive values are, are, are, are spreading. and yeah. so I, I, I think it's important to s- to make that connection. 

Yeah. No, I think...

That's why I really love that, chapter, was it, Billionaire, Class Consciousness, right? Where you, you lay this out. Because I think, there's still a lot of, misconceptions or shorthand stories that get in the way with understanding the post-pa- the pandemic and post-pandemic, developments, both in the market and in the market, a- and in the firms themselves, that also get in the way of understanding what came before, 

and like you said, how by virtue of, cultivating these progressive values, they, in anticipation, were like, "Okay, well let's, like, you know, let's crush the head of the snake." And then hardened, the anti-tech sentiment that led to, even more aggressive moves than I think they would've ever anticipated.

I don't think they ever would've anticipated, some Brandeisian, Brandeisian resurgence, for example, right? Or, a sustained effort at organizing firms internally or blo- and continuing to block them from doing certain lines of work. and by reframing so many of the tools that I think are just thought of as "Oh, this is just, the structure of a work.

You have pips, You have certain time, you have certain ways of doing time off, right? have, you have certain ways of, judging performance of a worker, certain ways of organizing the distribution of labor and contextualizing them to be like... No, actually these are all, like- Things that the tech companies explicitly adopted as ways to say "We gotta get this, we gotta get the mi- we pr- the, the woke mind virus out of our companies," essentially, right?

"We have to get this out so that we can get back to doing what we love to do," which is, m- fossil hydrocarbon contracts, fossil fuel contracts, the military contracts. protecting the homeland and its, its energy security, right? Among other things, of course. 

Yeah.

Yeah, exactly. I, I, I think it's, it's, this, this shift that we're talking about by the tech elites is something that, that they themselves talk about, you know- Yes ... in, in, in, in their circles, right? there, there are some tweets I, I, I remember that, I don't, I don't know if I'm gonna get this exactly correct, but, I remember in that period tech bosses telling each other just openly on Twitter, "You gotta fire the radicals to, to get things- Mm-hmm

under control," or, they would, they would... There's also a lot of name-calling, right? they would, they would say, "These are like, these t- these, these troublemakers, or these ESG grifters," right? Or, "They're, skittle-haired people with," a dig at at, at, at this group of workers.

Mm-hmm. 

M- Mark- Yeah ... Andreessen obviously is one of the, is one of the, he had, he had this big interview with The New York Times where he, where he openly talks about how his own radicalization was connected to, the tech worker movement, right? he was talking about how y- you could be a CEO of a company and get berated at an all hands meeting, right?

And, and these employees were so angry at you, just because you were, like, a white man. and so, it's, it's, it's so y- it's, it's so explicit sometimes in just how- Yeah ... in how they see things and, and h- how they talk about it. 

Yeah. this is something I j- And how it couldn't possibly be because of working conditions or, or because of how they were wielding their power in the workplace.

It, there's... It couldn't possibly be that. 

Y- yeah, admittedly a few years ago, this is something where I used to just be like, "Oh, these guys were always right-wing freaks. why are they talking as if they had a real road to Damascus moment?" The, but they did. They really did, some of them, where they thought that they were going on the middle path, right?

And, the resistance that they faced, as y- as you guys talking about, turned them even more into a right, rightward embrace, rightward turn. or maybe, and it also freed them to, I think, talk a little bit more about some of their, reactionary projects, but not just limit them to, the abstract.

political left and liberal, liberal elites out there, but, bring them back to the homeland, to their own companies, th- to introduce this sort of okay, we gotta do austerity, for the companies. We gotta really, yeah, s- like you said, fire the w- the radicals and brag about it online even though that's illegal 'cause who's gonna do anything, even if it's illegal.

Is the operating theory. I think o- o- one thing I would also be curious about is, as you, looking back and stepping back, do you guys think that looking at how they responded to the militancy, there was a way through where that was not the backlash? Or that contradiction was always going to just come up in one way or another, whether it was a movement that was less militant somehow, or a movement that was even more so and, and forced through, this sort of confrontation with, what they were doing and the values they'd cultivated and what they just wanted or needed as profit seekers, but also as, reactionaries in one way or another.

I, I guess I think that the, the, the rightward turn of the tech elites in particular, I, I don't wanna, I don't wanna sort of over-index on sort of the role of just the tech worker movement, right? I do think it is a, it is a very important part of the story. Yeah. But, this was also a period in which, in which, obviously there are broader political things happening in the world that had produced this kind of rightward shift as well.

Do you hate AI? The data's pretty clear right now, and it's intense. A March 2026 NBC News poll found only twenty-six percent of registered voters view AI positively, which means it's less popular than ice. Quinnipiac found fifty-five percent of Americans think AI will do more harm than good, up eleven points in a single year.

And Gallup found that only eighteen percent of Gen Z feels hopeful about it. Thirty-one percent of Gen Z say they're angry about AI. Apparently, we got a lot of AI haters out there. You might be one of them, or maybe you're just curious about what all the hate's about. Either way, we're gonna get to the bottom of it 'cause the anger isn't just showing up in the polls, it's now showing up in my mailbox.

I got this pamphlet sent to my house in Denver, paid for by a clean energy advocacy group. The front says, "A lot goes into data centers. What do we get out of it?" The inside lays it out with two big points. Data centers guzzle power, which means utilities keep aging coal plants running past their sunset dates and then pass those costs onto us.

And data centers also guzzle water for cooling. In a state like mine, where we're already facing water restrictions, this is a legitimate concern. Let's quickly address these two issues. I hear a lot about this water thing, but I think most people don't know the technology to fix the data center water problem already exists.

Microsoft is building closed loop systems that recycle water continuously without drawing from local supplies. And states like South Carolina and Kansas are both working on legislation to require this kind of closed loop system for any new data centers. This is a governance issue. Businesses won't just opt in to these extra costs.

We need to make it mandatory. We could, and we should. The coal problem has a similar fix. We just don't let data centers plug into our grid and leave customers holding the fossil fuel bill. We make them build new clean energy to cover the new capacity they need. And it turns out, Colorado already has a proposed bill for this.

It just needs to pass. So this fear-mongering rhetoric It's out there a lot. Some of it has a real basis in truth. But we can't let ourselves get sucked into the fear vortex. We gotta keep doing our research. That said, I do understand the fear and the anger. I've talked at length on this show about a bunch of really big challenges with AI.

But here's the other thing I am reasonably certain about: no matter how you or I feel about it, this train ain't stopping. This has been my position on this show since the beginning, and it's why I talk about AI as much as I do. If we're on this train, we need to understand how it works and where the frick it's taking us.

And if you're upset, that's reasonable. But if you vote a data center out of your town, tech companies will just build it in the next county or state or Canada or fricking space, apparently. The capital on this is too committed, and the momentum is too established. Getting angry at AI isn't gonna be a winning strategy.

Our anger needs a better target and, frankly, a better plan. I'm guessing you've heard the word Luddite. Basically a scared, backwards-looking, anti-technology peasant who can't handle progress. Well, that's what it means now. That's how I've always used it. But it wasn't always this way. Here's who the Luddites actually were.

It's 1811, England. The Industrial Revolution is underway. A group of textile workers in Nottingham begin a campaign of organized machine breaking, targeting the automated looms and knitting frames displacing their labor. They call themselves Luddites, followers of the mythical General Lud, a kind of Robin Hood figure for the industrial age.

Now, to this day, nobody knows if General Lud was a real person, but didn't really matter. He was a symbol of the idea that workers deserve a say in how transformative technology reshapes their lives. These were not ignorant peasants. They were skilled, trained, middle-class artisans, weavers who had spent years mastering their craft.

And what many people don't know is, before they broke a single machine, they tried every polite option available to them. In 1807, 130,000 workers signed a petition and sent it to Parliament demanding a minimum wage. 130,000 signatures. Parliament rejected it, mostly because the MPs voting it down were the same factory owners who would have had to pay the wage.

They asked for pensions for displaced workers. They asked for minimum labor standards. They requested that if machines were gonna take their jobs, the gains should at least be shared. They were ignored. So in March of 1811, they changed their methods and picked up some sledgehammers. They raided factories at night, targeting only the owners who were cutting wages and ignoring their demands.

In the early period alone, Luddite bands conducted at least 100 separate attacks, destroying roughly 1,000 machines. They were organized, strategic, surgical. The government's response was savage. Machine breaking was made a capital offense. At least 12,000 militiamen flooded the textile counties, and in a painful twist of irony, many of the troops were probably unemployed weavers.

In 1813- 17 Luddites were executed. Others were transported to Australia. The movement was crushed, not by argument, but by state force. And over time, Luddite got redefined from worker demanding fair terms in the face of displacement to idiot afraid of progress. Nobody even had to organize a smear campaign.

The word just got quietly repurposed during the computer revolution, and by the time it landed in our dictionary, the labor movement it came from had basically been written out of the definition. Since learning this, I've come to a rather strange realization: I think I'm a Luddite. Not in the redefined way, of course, but in the original sense.

I believe AI is transformative, it's happening, and in many ways, it's extraordinary. I just wanna know who gets the gains. I want the terms renegotiated. I want workers to have a seat at the table when decisions get made. That's the original Luddite position. Workers deserve a say in how transformative technology reshapes our lives.

one of the interesting things about how class consciousness started to crop up in the tech industry in the late 2010s was this moment where tech workers saw an opportunity to stand for the company's mission, right?

And they, stood up, said, "We should take this stance," and company leadership disagreed. That was a moment where workers realized that their interests were more aligned with the working class outside of their own companies than with their company's leadership, to begin with. And that kind of kicked off the tech worker movement, where people started to think more of themselves as workers.

They started to understand power structures in the workplace and build toward lasting change. 

That's fascinating. 'cause I was thinking, as JS was talking, what was the motivation here, right? if there's so much happening around 2018, we can think of a, a number of different things happening in that moment, right?

we're... You're not too long into the Trump presidency and, everything that that kind of brought with it. there's the controversy around, say, the Cambridge Analytic- Ana- Analytica revelations and how that is changing how people are seeing the tech industry and how these companies are operating.

we... As, as you're saying, Clarissa, challenging that narrative, but also this notion, of do no evil or, or whatnot that, was once associated with Google. Those things were being unraveled in that moment as well, right? y- you mentioned one reason for this organizing to be happening.

was there something broader that was going on as well, or, other things that were motivating this in that, in that time period? 

For a long time, tech workers, have been known for having some of the best jobs, right? Some of the cushiest jobs. Paid huge salaries, showered with perks. many of them even had the kind of, autonomy in their work that most other workers could only dream of.

And so, it, it is a bit of a, a, a puzzle as to why these, this group of workers ended up engaging in labor organizing. Like you were saying, Clarissa, I think it, it is super important to place this movement and the rise of this movement in the broader context in which it took place, right?

2018, the, the year the movement began, it was also the year, journalists and analysts were, were coining the term the techlash, right? It was this nationwide public backlash against the tech industry that, that was previously so celebrated, as not only the engine for innovation, but also for democracy and for, for being this bastion of progressive values.

But by 2018, right, a lot of these companies, a lot of these same tech companies started to get blamed for, for all kinds of societal issues, right? The spread of misinformation, the emergence of racially biased algorithms, the widening of political polarization, and I think, like you were mentioning, most of all, the election of Trump.

I think that was also the year in which the public broadly turned against the industry, and tech workers were very much in the middle of that, right? And they sided, in fact, with the public. People like Clarissa and I, when we were joining the tech industry in the 2010s, we very much believed that technology had this democratizing power and that Silicon Valley was gonna be the, the sort of perfect steward or vehicle for this v- for this vision.

And, y- I'll, I'll also say that, Silicon Valley was very much intentional in sort of marketing itself that way, right? They put on this progressive facade. They touted these grand missions about making the world more open, more connected, more democratic. Like Clarissa was saying, when, when these companies started to really get put under the pressure of capitalism and forced to profit maximize and build surveillance tools, weapons systems, exploitative, gig platforms and whatnot, it really became impossible to, to ignore how much the industry had gone back on these early promises.

And, and of course the workforce saw these abuses of corporate power as, as a major, major betrayal. By the time the, the techlash was in full blow, tech workers also were ready to mobilize. 

Yeah. And we, we see this over and over again in every stage of the tech worker movement, where these wider upswells of, resistance and transformative moments in society, like the Me Too movement, like Black Lives Matter, these are moments where tech workers feel strength in numbers from outside of their workplaces as well, which motivates them to stand up with, righteous, anger, and organize for a better society where they have the most leverage, which is at work.

Can you talk to me about that organizing as well, right? Because as you both mentioned, one thing that seemed really distinct when we think about the organizing in the tech industry was the issues that it was often associated with, right? when we think about unionization, we often think about pay, right?

what is the pay package that people are going to be getting to a lesser degree, do working conditions need to change? Does that need to be bargained? And certainly, that was, as I understand, part of the discussion, again, depending on the company that you're talking about. But there was also, this other kind of broader piece looking at societal issues that seem to be driving a lot of that organizing and, and getting a lot of those tech workers involved.

So can you talk a bit about the degree to which it was about those kind of, workplace issues versus those broader societal issues that the tech companies were very much implicated in when we think about climate change and militarization and, and those sorts of things? 

In the early tech worker movement, some of the most visible demands from tech workers were less related with what the labor movement terms bread and butter issues, like pay, hours, things like that, basic benefits.

and it was, it was more tied up with, how the technology was being wielded in society, or how the company was reproducing systems of oppression, like sexism. th- these were like larger issues in a way that went beyond the shop floor. but that being said, after those huge walkouts and amazing moments that catalyzed the tech worker movement, many tech workers that saw retaliation from taking part in these actions were more marginalized, were the more precarious workers at these companies, and that led tech workers to this idea that we need lasting structural power that will allow us to defend these massive wins that are on the horizon or that we've won already.

And that meant that a lot of the early union movement in the tech industry was led by the most precarious, most low-paid, most exploited workers at these companies, and Kickstarter is a great example of that. the, the trust and safety team, one of the lowest paid, hardest worked teams, underappreciated teams at Kickstarter- Of the people who worked that team and did that work, they were the ones leading the union drive, setting examples in how to organize and building power because they understood how important it was to balance the power in the workplace.

The characteristics of the tech worker movement where we say pushing back against the military, these really exciting radical things were also inextricably linked to bread and butter issues that would help empower workers to have the stability that they needed in order to engage in these struggles.

I think it's really important, like we were talking about, that there's so much resistance to AI in data centers that we also need to be thinking about what the alternative to this is instead of just saying, we need to stop the bad thing, which again, is important. What the AI build out shows, what the AI first agenda shows is that we can use, contrary to just like decades of propaganda basically saying the government is bad, we have to shrink the government.

That's a really concerted effort to, to benefit corporations and the wealthy, right? Which we are seeing with AI, but they're very clearly using the government so that-- So the whole thing is just a con at every level because they want to use the government to their ends. Some of what they're doing, at least the tools they're using for AI, this AI build out, are not necessarily bad.

Taking government stakes in companies, that could be a very useful tool for pro-social, pro-ecological ends. Like if we wanted to nationalize energy infrastructure and nationalize the energy grid, take stakes in companies to direct investment towards where we want to, to renewable energy, to public transit, et cetera.

We think it's really important to highlight that the project of world making should not be in the hands of the wealthy few, and that it should be in the hands of the working class, of the public, and the government is, is the main vehicle to do that. So, we have some recommendations like, uh, like public, public investment, democratically planning the grid.

Like instead of just letting big tech and AI companies just, just using the grid to facilitate their ends, like think-- using tools to, like the government to think about what is-- what should our energy grid look like? What should our energy system look like to benefit the planet and benefit people rather than AI companies?

Building a grid that is designed around social priorities, centering well-paying jobs, something like a green jobs guarantee. People talked a lot about a few years ago. I think we should-- need to be talking about that again because even if we set aside The sort of science fiction promises that AI is going to take everyone's job.

It is taking people's jobs right now, at least some people. I think it's a really important example to show that there's a lot of, there's a lot of precarity in our society that we need to use the government to, to prevent and ameliorate. And there's a lot of work that needs to be done if we're thinking about an energy transition, if we're thinking about repairing the biosphere and building a, something like a green and just society.

There's a lot of work that corporations are not going to want to pay for or pay well or provide jobs with dignity, and that's something we can use the government to do. And then AI, I think there's very clearly some uses or examples of AI That should be banned, right? When it comes to technologies and tech-technological advancement like this, we need public options and public control, public regulation, so it's not just Anthropic and xAI and OpenAI just doing kind of whatever they want.

and I, and I don't think that necessarily taking public ownership in those companies is, is even the best idea because I think we need from the ground up public alternatives that are fundamentally directed towards, social and ecological well-being. And who knows what that could look like.

We, we need public technology and computing infrastructure like that. And, I think that could really prevent and, mitigate a lot of the harms we're seeing. You can imagine what instead of an AI first agenda, we had a people and planet first agenda. And so think about that from like the t-the highest level where okay, how are we approaching the planning of our economy, the planning of our electricity grid, the planning of our-- the design of our cities and our land use that's actually benefiting well-being of ecosystems, the well-being of people first and foremost, and not, not Sam Altman or Elon Musk or whoever else, right?

So that's, that's the way we're, we're trying to, to think about it. 

Yeah. I think that makes a lot of sense, right? And, and just to pick up on a few pieces, like when you're talking about the, the jobs aspect of it, right? So many of these data centers promise a lot of jobs, but it's actually just construction jobs, and then there's not very much involved in actually running it, right?

the community is not actually getting, a lot of long-term employment out of these infrastructures that are very not just capital intensive, but resource intensive as you've been describing as well, right? And you mentioned the, the kind of democratic planning piece of this as well, and I feel when you talk about public technologies, obviously that's a way for us to get a lot more into or, or a lot more input on kind of the development of technologies, what technologies are being, put to use for the public rather than just whatever the tech industry is offering to us.

But it's also why I'm quite supportive of these efforts for moratoriums on data centers, right? It feels like that is an effort, however small or, maybe it's not as comprehensive as we would like, but at least to start to force some regulations, some expectations to start putting some rules in place around these data centers.

So it's not just the tech companies deciding for themselves what they wanna do and pushing the governments to do it, because the public is pushing on, whether it's their local governments or their state governments to get involved here, them having to put rules in place that, at least reflect some of these public demands to a certain degree in, what a data center, if it does get built locally, could actually look like, how it could operate, what resources it could use, and, and those sorts of things, right?

So much of this AI build out is very-- has been very undemocratic and even anti-democratic. I think we can think about moratoria as an exertion of democracy against that. This is happening so-- Like we've been talking about, this is happening so fast and to such a scale we have to slow it down, and we have to figure out alternatives to figure out what's going on and how we can, take that sort of power for alternatives and for democratic control 

No, I, I completely agree.

And, as we start to wrap up our conversation, I, I'm wondering about the path forward, right? obviously there are a lot of forces arrayed against us from the tech industry and their lobby and, everything that you've been talking about on that front. But even, as you've been saying, governments that are not always acting in our interest as well, right?

And are supporting what these tech, companies want. And so I wonder for you, what you see as, the path toward moving toward a framework like the one that you're laying out and, what kind of hints or, or, kind of movements in that direction you're seeing right now that can help us along, in that fight or, or in the achievement of that.

The future is up for grabs right now, and visions of the future, there's not really a-- it doesn't feel like there's a lot on offer. Like I was saying earlier, it feels like the AI future is very vague, and it's also doesn't sound good really. even, even like in their most like, their most lofty promises, like that doesn't really sound good to me.

and I think, it doesn't sound good to a lot of people. So I think there's a lot of space, and that has to be contested for like what could an alternative future look like. That's where people's needs are met, where people are not, co- people's cost of living is not going through the roof, where corporations aren't allowed to pollute with impunity, where we have renewable energy and public transit and libraries and schools and, and all these sorts of things that I think people want and people need.

What we're seeing with this data center resistance is a very interesting and potent sort of political moment of, of potential, of where people are fighting against these sort of undemocratic oligarchic forces. And maybe they're not always, they're not always like someone who's coming at that from like a, a traditional left politics perspective, but that's a, that's a place where you can say, "Hey This data center that, that is being imposed on you, well, here's who's doing that and here's why, and here's, the negative effects, and, here's what maybe what some alternatives to be-- could be.

And maybe that's, maybe the groups organizing a data center resistance could organize to-- for something more beneficial in their community, right? So to me, that's like a place of political potential is this data center resistance, and we're seeing a lot of candidates running on that. I think-- I don't wanna, sugarcoat it.

I think it's gonna be a big challenge because as you talk about, as we've been talking about this conversation, the inequality in our, in our society is just so enormous right now from Elon Musk briefly being the world's first trillionaire to, increasing cost of living and, to the working class and the way that the Trump administration has just been so blazen- brazenly corrupt and authoritarian.

So it's a big challenge, but I do think that with the way that they've been pushing this so hard, I think it's, it's a potential-- there's a lot of potential for political openings here. And I think it also shows that contrary to maybe some-- what some people are saying right now or what we're seeing from the Trump administration's actions, that people do care a lot about the environment and, and climate change and that, I, I think That's one of the things people are citing the most from what I've seen with why they're resisting data centers, is concerns about the environment.

By doing this, they're drawing attention to it, and, I think that that's a lot of space for, for us to, to collectively imagine alternatives. 

"What does a complex global civilization require to function?"

Because it's that complex adaptive system that allows the ultra wealthy to live their ultra lavish lifestyles. And when we answer the question of what a complex global civilization needs to function, the logic of this solution breaks down pretty quickly because the ultra wealthy depend on systems that require a lot of other people.

Think about what it takes to maintain a billionaire lifestyle: private jets that require global supply chains for fuel and parts and pilots and air traffic control, luxury goods that require designers and craftspeople and logistic networks, medical care that requires doctors and researchers and pharmaceutical supply chains, technology that requires engineers and chip fabs and rare earth mining and electricity grids.

You couldn't run modern civilization with 100 people and robots. Feudal lords had all the land and weapons, but they couldn't maintain Roman-level complexity with tiny elite populations. Why? 'Cause complex systems require large populations for specialization and redundancy and innovation. Without those things, the system simplifies.

Less trade, less coordination, less resilience. Modern billionaires would face the same constraint. They might survive, but they wouldn't be able to maintain current levels of complexity. Complex systems need the conditions that sustain complexity. Remove billions of people, you don't get the same civilization.

You get a simpler one. So I don't believe there's any depopulation plan, but I do think there's a trajectory where capital optimizes for fewer workers, more wealth concentration, and further isolation of elites. Not because anyone's planning it specifically, but because that's where the current incentives lead.

It's not inevitable, it's just the currently incentivized trajectory. We talked about this last week, and here's the good news. Trajectories, wealth only means something in the container of a functioning society. Money is a social technology It requires collective belief. If society collapses, their billions are just numbers in a database that nobody gives a crap about.

They need us more than they think, and we have more power than they seem to remember. So those are the main varieties of crises that people seem to be imagining. None of those possibilities quite hold up under scrutiny, but all of our collective intuition, I think, still seems to be screaming, "There is a crisis coming, and AI will have something to do with it."

I will admit, my spidey sense says the same thing. At a macro level, humanity has seen this pattern before, multiple times across thousands of years in every major civilization, I think. Our societies organize and then reorganize over and over. The word collapse often gets invoked in these situations, but to me, that word implies something that isn't exactly happening, which is why I like this notion of reorganization better.

We've talked about this transition before in relation to the fourth turning. There's a crisis and then a reorganization, a winter and then a spring. So when people ask, "Can we prevent an AI crisis?" I can tell you what history suggests. Probably not. The reforms we desperately need, in this case, income replacement, automation dividends, public AI stakes, new social contracts, these kinds of things don't tend to happen before the disruption.

They happen after. After the Triangle Fire, after the suffragists actually suffer, after the Great Depression, after the organizing conflict forces enough of us to choose. So if we can't preempt the crisis, what is the point of me telling you all of this? Why do I make these episodes week after week? It's because of something that I call the chaos window.

The chaos window is the period between when the old system breaks down and the new system takes hold. It's the messy, melty transition from winter to spring. It's when society experiences maximum uncertainty and often maximum pain. But it's also the season that contains maximum possibility. And here's what history shows.

Some chaos windows are short and some are devastatingly long. But what makes them different isn't luck, it's preparation. And we're gonna go into this in more depth in next week's episode, but for now, I'm gonna summarize with a contrasting example from Chinese dynasties. After the Han dynasty collapsed in the year two twenty, China fragmented for three hundred and seventy years before the Tang dynasty reunified it.

After the Tang collapsed in the year nine oh seven, it only took fifty years before the Song dynasty pulled it back together. Same pattern, same civilization, but one chaos window was seven times longer. What was the difference? After Tang, they had recent memory of unified governance. They had models, infrastructure, shared frameworks.

They were prepared to rebuild. So they didn't spend centuries wandering. They spent fifty years. That's three hundred and twenty years of suffering avoided. Actions of the leaders in the Song dynasty didn't prevent the collapse, and in many ways, I don't think that's the goal anyway. The crisis is what allows for the change, just like the decay of winter is what allows for the bloom of spring.

We can't avoid winter, but we can help determine how long it lasts. The leaders of the Song dynasty did shorten the chaos window, and that's why this work matters. Not because we're gonna prevent the AI crisis, whatever it might look like, but because we're preloading the solutions. We're building vocabulary.

We're circulating frameworks and planting ideas. So when our chaos window opens, when the old system visibly breaks and everyone's asking, "What now?" Some ideas are already there. Will they be perfect? No, but they will be ready to start testing. Is it noble to prevent suffering? Yes. Is it more noble to shorten the duration of suffering, even if you can't prevent it?

Maybe. Because every year we shorten the chaos window is a year of fewer families destroyed and fewer lives derailed and less generational trauma created and a faster recovery for our overall society. This isn't small work. It is not for the faint of heart. It is civilization-level work, and it really fricking matters.

We can do this, my friends. If you're still here in this community saying hello tomorrow with me, I'm guessing this might get you a little fired up too. So this week's optimistic rebellion is to look back in order to look forward. Think about the New Deal. It didn't appear from nowhere in nineteen thirty-three.

Those ideas, progressive taxation, social insurance, labor protections, they've been circulating for decades. They've been simmering quietly, gradually, until ready for the suddenly. When the Great Depression hit, the frameworks were ready to deploy. That shortened the chaos window. Same principle here. Who buys your stuff, robots?

It plants a question. The Fourth Turning framework gives us context. Economic systems thinking provides us with vocabulary. Every week, we are understanding the problems more and more clearly, and this allows us all to start brainstorming solutions now. And all of this will help us shorten our chaos window when it opens.

So when someone asks, "Can we prevent an AI crisis?" We know the answer's probably no, but we also know the exact future isn't inevitable. There are still so many choices to be made between here and there, choices that will change the direction of the future we will actually get to experience. 

going to be it for today.

As always, keep the comments coming in.

You can record - and re-record - a voice message by tapping the link in the show notes,

You can reach us on Signal at the handle bestoftheleft.01,

or simply email me to [email protected]

The additional sections of the show included clips from;

Pivot to AI

Blood in the Machine

Michael Burns

Democracy Now!

struthless

Harland Spinks

Marcus Hutchins

Owen Jones

The Lever

The Tea with Myriam François

PissedMagistus

Side Burns

This Machine Kills

Hello Tomorrow

and Tech Won't Save Us

Further details are in the show notes.

Thanks to everyone for listening, thanks to Deon and Erin for their production work for the show, thanks to Amanda for all of her work behind the scenes, thanks to our editors and thanks to those who already support the show by becoming a member, purchasing gift memberships, or making one-time donations.

You'll find the link to support us in the show notes along with links to join our Patreon and Discord communities for free where you can also continue the discussion. And don't forget to follow us on all the social media platforms as I prepare to relaunch our social media strategy!

So coming to you from far outside the conventional wisdom of Washington, DC, my name is Jay! And this has been the Best of the Left podcast, coming to you twice weekly, thanks entirely to the members and donors to the show, from bestoftheleft.com.