Etharchy All Rabbit Holes

🕳️ Down the Rabbit Hole

The Perfect Scapegoat

AI is coming for your job, and then it's coming for you. That's the story, anyway. But a tool doesn't decide who it serves — people do. So who keeps telling us to fear AI, and who walks away while we do?

ai power labor democracy

Artificial intelligence is coming for your job. Then it’s coming for you.

That’s the story, anyway. In September 2026, a researcher at one of the world’s leading AI companies quit and warned that the machines he’d helped build could slip out of human control. A colleague who stayed put the odds of AI ending humanity within a decade at better than one in ten. That same year, college graduates booed commencement speakers for so much as mentioning AI, while the entry-level jobs they’d trained for quietly disappeared. Fear of AI had never been higher.

But every age has its monster under the bed. And it’s always worth asking who keeps putting it there.

A fair warning before we go on: what follows is not a popular opinion. Right now, nearly everyone agrees AI is the thing to fear — left and right, workers and pundits, the people building AI and the people protesting it. We’re going to argue that it isn’t. So we’ll ask one favor: read to the end before you decide. You may still disagree. But you’ll know exactly what you’re disagreeing with.

The tool and the hand

Here’s the thing about machines: they don’t want anything. An engine is just a very clever furnace, and nobody blames the furnace for where the car goes. AI is the same. It’s a remarkable tool, maybe the most powerful one we’ve ever built, but it’s still a tool. Someone decides what it’s pointed at. Someone decides who it replaces, who it helps, and who keeps the money it saves.

The masters of mankind have always understood something the rest of us keep forgetting: when things go wrong, the most valuable thing you can own is someone else to blame. For centuries, that meant a group of people — immigrants, the poor, a minority, whoever was useful that year. Those targets haven’t gone anywhere. We’re still handed them every election cycle. But now there’s a new arrow in the quiver, and it may be the most convenient one yet.

It isn’t a someone at all.

Nobody defends a machine

Think about why that matters. When a group of people gets blamed, someone eventually stands up for them. Advocates, families, churches, lawyers, neighbors — people who know them and refuse to let the lie stand. It’s slow, and it costs dearly, but the truth fights back.

Who stands up for a machine? Only the people selling it. And that’s the trick. When a billionaire stands at a podium and tells you AI is the future and you’d better get used to it, every instinct you have says don’t trust this man. So you push back — against the tool he’s praising. The more the masters defend it, the more we hate it. Their defense doesn’t protect AI. It protects them.

And there’s no guilt in it. History’s worst crimes began with blaming people, and somewhere in the back of our minds we know where that road leads. Blaming a machine feels harmless. Nobody gets hurt. That’s exactly what makes it so effective: the anger goes somewhere, it feels righteous, and the people who actually made the decisions never even have to step out of the way.

The graduates

In May 2026, at college graduations across the country, commencement speakers stood up to talk about AI — and graduates booed them. At the University of Arizona, they booed a former Google CEO. At the University of Central Florida, they booed a speaker who called AI the next industrial revolution. At Middle Tennessee State, a record-label CEO answered the boos by telling the graduates to “deal with it.”

Their anger was real, and it was earned — and it wasn’t only theirs. In 2025, American employers announced more than 1.2 million job cuts, the most since the pandemic year of 2020. It wasn’t just tech. Government, retail, warehousing, transportation, health care — the cuts reached nearly every corner of the economy, and they kept coming into 2026. At the same time, the door to the next job was closing. Hiring plans in 2025 fell to their lowest level since 2010. Recruiters were wading through nearly three times as many applications per hire as a few years earlier, and automated filters were rejecting people in minutes. Fresh graduates found the entry-level door shut. Experienced professionals with decades behind them sent thousands of applications into silence. If you worked for a living, it didn’t much matter whether you had two years of experience or twenty.

But look at what the graduates booed. The boos landed on AI, not on the layoffs or the hiring freeze. People just like the ones who made those decisions — to cut at every level, stop hiring at every level, and spend the savings on data centers — were standing right there at the podium, and they walked off to polite applause.

Maybe it’s a plan. Maybe it’s just luck. Either way, follow the money, and ask who it works out for.

Why blame AI?

So why blame AI at all?

Because it’s a better story. “We’re cutting people because it makes our numbers look better this quarter” is not something any company will ever say out loud. “We’re embracing artificial intelligence” sounds like the future. Investors reward that story. For much of 2026, AI was the reason companies gave most often for their layoffs. By October, the plain old explanation — “market and economic conditions” — had quietly overtaken it for the year.

And look at who was being shown the door. The engineers who built the cloud platforms, wrote the code, and laid the foundation every AI system now runs on. Much of what today’s AI knows about writing software, it learned from code those same workers wrote. They weren’t surplus. They built the tool. And then the tool got the credit for replacing them.

The evidence had been piling up all along. In the fall of 2025, economists at Yale looked at the whole American job market in the nearly three years since ChatGPT arrived and found no discernible disruption. In January 2026, researchers at Forrester found that many companies announcing AI layoffs didn’t have AI anywhere near ready to do those jobs. Even the head of the world’s most famous AI company admitted that some businesses were blaming AI for cuts they would have made anyway.

There’s a name for it now: AI-washing. It’s the oldest trick there is, with a fresh coat of paint.

The words

Before we can talk about what we’re supposed to be afraid of, we have to talk about the words.

The old dream of a thinking machine was one machine that could do what people do: drive a car, play chess, bake a cake, change a lightbulb, fold the laundry. Steve Wozniak once boiled it down to a simple test: could a machine walk into an ordinary home it had never seen and make a cup of coffee? Despite plenty of impressive demos, no machine can reliably do that today. Yet one leading company’s charter defines artificial general intelligence as systems that outperform humans “at most economically valuable work,” and others have skipped ahead to talking about “superintelligence.” In practice, the physical world — the part machines are worst at — quietly fell out of the definition. When the finish line keeps moving, anyone can announce they’ve crossed it, or that they’re about to. Either way, the announcement is worth money.

So why keep moving the bar? Notice that the word gets used two ways, depending on what’s needed. When it’s time to raise money, AGI means something modest: software that can do most office work, and look how close we are. When it’s time to frighten us, AGI means something out of the movies: a mind with its own goals that might turn on us. The same word, sliding back and forth. And the movie version does one more job. A tool can’t be blamed for anything. A mind can. Call the tool “intelligent,” hint that it might be thinking, and a tool becomes a someone: someone to blame, but no one to defend.

The fear

Then there’s the fear itself. Remember the number from the beginning — better than one in ten that AI ends humanity within a decade? Ask where it comes from. When a doctor tells you a disease has a ten percent death rate, that number comes from counting: a thousand patients, a hundred deaths. When analysts forecast the odds of a government collapsing, they build models from decades of real collapses. Every real forecast works the same way. The thing being predicted has happened before, over and over, and you can learn from every one.

Human extinction has never happened, but that alone isn’t the problem. Asteroids have never wiped us out either, and we can still estimate those odds: we can count the craters, and we’ve watched rocks slam into other planets. We can estimate the danger of nuclear war because we know exactly what the weapons do. We’ve seen them used. The danger is real, and so is the mechanism.

AI has no mechanism to count. No machine has ever shown anything close to the ability to overpower humanity, and nobody has shown that one can be built. Every doomsday scenario is a chain of steps, each one theoretical. “One in ten” isn’t a measurement. It’s a guess about something no one has shown is possible, dressed in the clothing of math.

Ask the researchers themselves and the answers are all over the map. In one large survey of machine-learning researchers, the typical guess for extinction was about five percent, while nearly one in ten put the odds of an extremely bad outcome at fifty percent or higher. That’s not a finding. That’s a room full of people who don’t know. Not knowing is honest. Selling the guess as a certainty is not.

So who’s selling it, and why? Look at who benefits. A technology that might end the world is a technology only a few trusted companies should be allowed to build. The companies warning loudest about AI’s dangers are also asking to be regulated — and rules written with the help of the biggest players tend to be rules only the biggest players can afford. The fear doesn’t slow them down. It pulls the ladder up behind them.

The race

And whenever the fear comes up, listen for what comes right after it: and that’s why we have to beat China.

Here’s what rarely makes the headlines. China’s population has been shrinking since 2022. Its birth rate is among the lowest in the world, and its workforce is aging fast. Much of its AI push is aimed at something practical: factories, logistics, elder care, robots to do the work of people who will no longer be there. China is no saint in this — it has built one of the most extensive AI surveillance systems on Earth. But “a doomsday race for a superweapon” is a story, and it’s worth noticing when that story gets told. It tends to show up right before a request.

In November 2025, the chief financial officer of the world’s most famous AI company floated the idea of a federal “backstop” for the industry’s enormous infrastructure debts, pointing to the race with China. After a public outcry, she walked it back. Weeks later, the White House’s top AI adviser posted that AI investment accounted for half of America’s economic growth, and that pulling back would risk a recession. By the spring of 2026, the financial experts surveyed by the Federal Reserve had moved AI up to third on their list of threats to the nation’s financial stability.

If this sounds familiar, it should. In the 1870s, it was railroads — a story we’ve told before: a world-changing technology, everyone with money pouring capital into it, a bubble, and a crash that fell hardest on people who never owned a share. The masters of mankind have always loved this arrangement. When the bet pays, the profits are private. When it doesn’t, the losses become everyone’s problem.

So when you hear “we have to win the race,” ask the next question: who’s paying for the race, and who keeps the prize?

The tool itself

Here’s what gets lost in all the noise: AI is a good thing.

It has already predicted the structures of hundreds of millions of proteins, a puzzle biologists had been stuck on for fifty years. It has helped discover new antibiotics. It reads medical scans, forecasts storms, and helps keep the power grid running. And this is the beginning, not the end.

Every great tool in human history has looked, at first, like a destroyer of work. The printing press put scribes out of business. The automobile ended the livery stable and the blacksmith’s shop. Each time, the fear was real, and the transition was hard on the people caught in it. And each time, the tool ended up creating far more work — and far better lives — than it destroyed.

We believe AI can do the same, and more. It could make medicine cheaper, energy cleaner, and the hard, dull parts of work disappear. It could raise the quality of life for ordinary people more than anything since electricity. But a tool doesn’t decide who it serves. People do.

And here’s the cost of all this fear, the one nobody mentions. Every story about killer machines and stolen jobs teaches us to keep our distance from the tool itself. Think about the logic: if AI could become powerful enough to end humanity, then it’s powerful enough to change humanity for the better. Neither is promised. Both are “what if.” But only one of those what-ifs gets the headlines.

Ask yourself: if the people holding the most powerful tool ever built keep teaching the rest of us to fear it, who ends up deciding what it’s for? Not us. We’ll be standing at a distance, wary and suspicious, while they choose for us.

So the answer isn’t to fear AI, or to smash it. It’s to embrace it — and to make sure it’s aimed at bettering humanity rather than further enriching the masters of mankind. That won’t happen on its own. It never has.

Nobody asked us

So who’s really to blame? It’s tempting to say a few greedy people, and some of them have earned it. But look closer. An executive who refuses to cut jobs while his competitors do can be punished by his shareholders, or replaced by someone who will. The system doesn’t just tolerate the vile maxim. It pays for it. And systems aren’t weather. People built this one, and people keep it running.

Between 1811 and 1816, English textile workers broke into factories and smashed the new machines. History calls them Luddites and uses their name as an insult for anyone afraid of technology. But many of them knew those machines better than anyone. What they fought wasn’t the machine. It was how the machine was being used: to cut their wages, break their trades, and hand every gain to the owners. Parliament’s answer was to make machine-breaking punishable by death. The workers themselves never got a vote. Most of them couldn’t vote at all.

We’re supposed to be different. We live in a democracy. The people are supposed to get a say, especially on the big things. And according to the people building it, this is the biggest thing there has ever been.

So which is it? If AI really carries a one-in-ten chance of ending humanity, then it’s the most important decision our species has ever made — and it’s being made in private, by a handful of company boards. And if it doesn’t, why are they telling us it does?

Either way, nobody asked us.

This machine is not the enemy. It never was. It is a tool — maybe the best one we have ever built — and the only question that has ever mattered about a tool is whose hand is on it.

The enemy is the other machine, the one that has always run on a single rule: all for ourselves, and nothing for anyone else. Right now, its hand is on the tool.

It does not have to stay that way.

Sources

The opening

The graduates

Why blame AI?

The words

The fear

The race

The tool itself

Nobody asked us

← More Rabbit Holes Back to the book →

Found this worth reading? Pass it along.

Download the book