You know that feeling when your phone buzzes and you jump? Not because it’s your partner, or your kid, or even your boss—but because the app told you to move, now. That’s not productivity. That’s servitude.
Cory Doctorow calls it the reverse centaur: a human body strapped to a machine that doesn’t care if you live or die, only if you comply. It’s not the robot doing the job—it’s the human, trembling, trying to keep up with an algorithm that’s never slept, never got sick, and never had a bad day. And when the algorithm fails? You’re the one who gets fired.
Amazon delivery drivers are the poster children. Cameras on their vans track every second of their movements. Speed? Too slow. Too fast? You’re a hazard. Did you take a bathroom break? That’s a violation. The AI doesn’t care if you’re pregnant, if your kid’s fever spiked, if your back gave out. It just wants the package delivered on time. And when it misreads a street sign? You’re the one who gets blamed. Not the algorithm. Not the vendor. Not the engineer who coded the vision system. You.
It’s the same in hospitals. Nine radiologists fired. One left. Her job? To verify the AI’s diagnoses. Not to interpret. Not to think. To rubber-stamp. And if a tumor gets missed? She’s on the hook. The AI didn’t make a mistake—it was just following its training data. The human? She’s liable. That’s not augmentation. That’s exploitation dressed up as efficiency.
This isn’t science fiction. It’s Tuesday.
$1.4 Trillion in the Air—And Nothing to Show for It
Let’s talk money. Real money. The kind that makes CFOs salivate and investors weep with joy.
Global AI capital spending? $1.4 trillion. That’s not a typo. That’s more than the GDP of Switzerland, and nearly half the entire annual U.S. defense budget. Meta alone spent $150 billion over three years. And they’re not done. They’re planning another $150 billion this year. That’s not investment. That’s gambling with the economy’s life savings.
And what’s the return?
About $50 billion in revenue. Per year.
Let that sink in. You’re losing $1.35 trillion for every dollar you make.
The web didn’t work like this. When you used the web, you made money for the company. Every click, every search, every purchase—profits climbed. AI is the opposite. Every time you use it, the company loses more. Not a little. A lot. Each new model is more expensive to train. More power-hungry. More fragile. And it still hallucinates. It still fails. It still produces garbage. But you pay more to get less.
It’s like buying a Ferrari that only moves when you kick it, and then it breaks down after five miles. And you’re paying $200,000 for the privilege.
And yet, the money keeps flowing. Why?
Because the capital markets have the object permanence of a toddler.
The Fantasy of a World Without People
Here’s the real secret. The one they don’t put in the investor decks.
AI isn’t being built to make things better.
It’s being built to make people disappear.
Think about it. Why do corporate leaders love this stuff so much? Not because it’s efficient. Not because it’s smart. Because it solves the problem of other people.
Hell is other people, right? We all know it. But for the people at the top? It’s worse. They’re haunted by the quiet truth: if they don’t show up for work, the company still runs. If their employees don’t show up? Everything stops.
AI fixes that.
You don’t need to negotiate with a union. You don’t need to listen to complaints. You don’t need to pay overtime, or offer healthcare, or deal with someone who says, “I’m tired.” You just type something into a chatbot—and it shits out your product.
That’s the fantasy. A business without workers. A government without employees. A world where power doesn’t have to negotiate with humanity.
And it’s not just tech. Look at the government layoffs. Look at the public sector cuts. Look at the way AI is being sold as the solution to “bureaucratic inefficiency.” The real inefficiency? People asking for raises. People demanding dignity. People having lives.
AI doesn’t have a life. So it doesn’t need one.
The Bubble Is Built on IOUs
Every bubble has a story. The dot-com bubble? “We’ll monetize eyeballs.” The housing bubble? “Homes always go up.” The crypto bubble? “Decentralization.”
The AI bubble? “We’ll lose money until we don’t have to.”
Seven companies account for more than a third of the entire stock market. And they’re all passing around the same $100 billion IOU. They’re not building products. They’re building narratives. They’re not creating value. They’re creating illusions.
And here’s the kicker: every AI model they train? It’s obsolete in 24 to 30 months. The hardware rots. The software decays. The data becomes stale. And the energy cost? Astronomical.
When the money dries up—and it will, because no one can sustain $1.4 trillion in losses forever—the data centers will shut down. The models will vanish. The startups will collapse. The engineers will be laid off. And the companies that bet everything on AI? They’ll be left with nothing but debt, lawsuits, and a generation of workers who spent their careers serving machines that never learned to care.
This isn’t a correction. It’s a collapse.
The Asbestos in the Walls
Doctorow calls AI “the asbestos in the walls of our technological society.”
Think about asbestos. It was sold as a miracle. Fireproof. Cheap. Perfect. Everyone used it. And then, decades later, people started dying. Not from the fire. From the dust. From the hidden poison.
AI is the same. It’s not dangerous because it’s sentient. It’s dangerous because it’s everywhere. And no one’s going to tear down the walls.
We’ll keep using it. We’ll keep relying on it. We’ll keep blaming the humans when it fails. And we’ll spend the next 20 years cleaning up the mess.
Because once you embed a technology into the infrastructure of society, you don’t just unplug it. You have to excavate it.
And that’s what we’re facing.
We’re Not All Reverse Centaurs
Here’s the thing: not everyone’s a reverse centaur.
Some people—real people, not corporate PR bots—are using AI as a tool. A real tool. Not a replacement. A helper. A way to draft emails faster. To summarize reports. To translate documents. To brainstorm.
They’re the centaurs. The ones who control the machine. Who decide when to use it. Who know its limits. Who don’t let it make their decisions.
But they’re the minority. And they’re being drowned out.
The system doesn’t want centaurs. It wants reverse centaurs. Because centaurs have agency. Reverse centaurs? They’re disposable.
And that’s why the bubble is so dangerous. It’s not just about money. It’s about power. It’s about who gets to decide what work is worth doing. And who gets to decide who gets to do it.
We’re not just building a new economy.
We’re building a new kind of slavery.
And we’re the ones holding the leash.
The Only Way Out Is to Break the Machine
There’s no graceful exit here.
You can’t tweak the algorithm. You can’t regulate the data centers. You can’t ask nicely.
The only way to burst this bubble is to stop feeding it.
That means refusing to use AI where it replaces human judgment. Refusing to sign contracts that force you to use it. Refusing to let it be the default.
It means supporting the workers who say, “No.”
It means defending the right to scrape the web—not to train models, but to preserve history. Because if we lose that, we lose everything.
It means demanding that AI spending be treated like nuclear waste: not an investment, but a liability.
And it means remembering this: the web didn’t become profitable because it was unprofitable. It became profitable because people used it to do things that made money.
AI doesn’t do that. It doesn’t create. It consumes.
We’re not waiting for the bubble to pop.
We’re waiting for someone to finally say: enough.
