The Misalignment Is Human
AI eats the coordination layer first — the measurers go before anyone else. The danger that follows isn't a rogue machine. It's us.
Unsafe Intelligence
The layoffs aren't coming. They're here. Meta cut roughly a tenth of the company and moved most of them onto AI teams. Cloudflare cut a fifth of its workforce — its first mass layoff ever — in the same quarter it posted record revenue. And the CEO said the quiet part out loud in the Wall Street Journal: the people he replaced were "measurers" — middle management, finance, legal, internal audit. The connective tissue. The layer that exists to coordinate the other layers.
To see why the middle goes first, you have to see that a company carries two different kinds of complexity.
The two complexities
Product complexity is the genuinely hard part — serving billions of users, the uptime, the abuse, the edge cases, the legal surface area. That doesn't go away, and it never will.
Human-to-human complexity is the other kind: the coordination tax. Every person you add is a new node on the social graph — their goals, their incentives, the meetings about meetings, the emails to be cc'd on. Anyone who has worked at a big company knows the productivity per head past a certain size is embarrassing. The old wisdom was that this tax was the unavoidable cost of scale. You wanted to grow, you ate it.
AI's biggest short-term leverage isn't curing cancer. It's eating that coordination tax. So the measurers go first — and the leverage of the people who remain goes vertical. The 10× engineer becomes the 1000× engineer; a fifty-person company books tens of millions in revenue per head; the first one-person billion-dollar company stops being a thought experiment. Fewer people holding far more. That is the defining move of the next few years, and it is already underway.
Now the part nobody wants to think about
Here is the turn. The AI-safety conversation has spent a decade worried that the machine will be misaligned with human values. The nearer, likelier danger is the reverse: that we will be misaligned with each other.
Picture a large, growing pool of newly displaced, formerly six-figure professionals — people whose entire identity was built around being the smart, well-compensated one — competing for fewer, lower-paid roles, and watching the people who automated them become some of the richest humans who have ever lived, on television. That is not a stable arrangement. And the resentment it produces will be real, and it will be earned.
The psychology isn't a mystery. People feel losses about twice as hard as equivalent gains. They measure their lot against the person one rung up, not against history — last-place aversion is a documented thing. And it's status threat, more than raw economic hardship, that reliably tips people toward political revolt. The velocity of the status reversal is what makes it toxic — people can absorb an economic shock; they cannot absorb one while the cause of it is being celebrated.
We have run this experiment before. The Luddites weren't technophobes — they were skilled workers whose wages were being cut, and in 1813 seventeen of them were hanged at York. The Captain Swing rioters smashed threshing machines across England in 1830; the courts handed down 252 death sentences. And the modern data rhymes: after financial crises, far-right vote share rises about 30% on average. Dislocation doesn't stay economic. It becomes political, fast.
The real alignment problem
So the thing to fear in the next five years isn't a superintelligence deciding it doesn't need us. It's a few million justifiably furious people optimizing locally — for themselves, right now — and burning the whole project down before its gains are ever shared. Regulation that slows only us. Capital flight. Sabotage. The pitchforks.
That's not an argument against AI. It's an argument that the binding risk is social, not technical — and that the people building this should be spending at least as much worry on the humans in the trough as on the model in the datacenter. The hardest alignment problem of the decade doesn't have a loss function. It has a face, and it looks like ours.
We are not afraid of the machine. We are afraid of us. Which is exactly why the way through has two moves, not one — cross the middle fast, and share the gains before they're taken. But that's another essay.