The most powerful general-purpose technology in human history has arrived and the best idea anyone can come up with is that we should all become plumbers.

The pessimists say we’ll be unemployed. The optimists say we can retrain into trades that AI can’t reach yet — electricians, mechanics, welders — until robotics catches up, at which point presumably we’re all out of ideas and will be living on some kind of welfare in a post-scarcity economy. We spent a century building an economy where fewer people had to break their backs in physical labor. That was progress. That was the point. And now we’re told the silver lining is manual work. The whole area of thought is ridiculous, and the fact that serious people entertain it tells you more about the poverty of our collective imagination than about the nature of the technology.

Both camps share the same broken assumption: that the total quantity of work in the world is fixed. Economists call this the lump of labor fallacy. It’s a well-known error that intelligent people become peculiarly susceptible to the moment fear enters the picture.

The Failure of Imagination

Here’s what actually happens when you give a company AI.

The naive version — the one most executives reach for because cost-cutting is the only instrument they’ve ever learned to play — runs like this: ten people produce output X. Fire five, give AI to the remaining five, maintain X. Headcount down, margins up, board is happy.

But why stop there? Keep all ten. Give everyone AI. Now each person does what used to require an entire team. Output doesn’t stay at X. It goes to a hundred times X. The return on each employee becomes so extraordinary that the rational move isn’t to fire people — it’s to hire more of them at this new, almost absurd level of productivity.

If one person with AI can match the output of a pre-AI company, then a hundred-person company should produce the output of a hundred companies. This isn’t fantasy. It’s arithmetic.

The Twitter Lesson

Companies are firing people and blaming AI. I don’t buy it.

Elon Musk showed you could run Twitter with maybe ten percent of the workforce. The platform kept running. What were the other ninety percent doing? This isn’t an AI story. It’s a management story — organizational bloat, and the extraordinary capacity of large institutions to absorb talent without producing commensurate output.

Here’s the counterfactual nobody considers. If each of those fired employees had been equipped with AI and pointed at real problems, it might never have made sense to let them go. Each one producing the value of the ten percent who stayed. That’s nine times the output. Imagine where the platform would be now. Imagine everything they could have built. Instead they were fired because nobody in authority had the imagination to hand them a force multiplier and aim them at something bigger.

Why AI Doesn’t Replace Us

The objection shifts: maybe AI replaces humans entirely. This is confused.

Someone has to be held responsible. That’s not a negotiable feature of civilization — it’s its foundation. The world isn’t a clean optimization problem. We try to kill each other. There’s geopolitics and lobbying and lust and ego and religion and terrorism and the butterfly effect that makes every person and every situation irreducibly unique. A million factors influence what we want and why we want it. Human beings need to be involved with all of them, will disagree about all of them, and must be accountable for the decisions that result.

The right analogy is mythological. In every civilization’s mythology the gods controlled the fate of mortals — they directed, they judged, they decided. The mortals served. That’s the correct arrangement. We’re the gods in this picture. AI agents are the mortals who execute our will. Ceding that authority isn’t something any sane civilization would do.

The Promotion of the Species

What follows isn’t that humans become irrelevant. It’s that human work transforms. The right mental model isn’t the replacement of soldiers — it’s the promotion of an entire army to the rank of general.

Let’s be honest about what AI is actually replacing. Horrible work. Mind-numbing grunt work that turns intelligent people into expensive automatons. The great tragedy of the modern economy isn’t that work is being automated — it’s how much human potential we’ve been incinerating on tasks that never required a human being in the first place. AI doesn’t take the good parts of your job. It takes the parts you complain about at dinner.

Everyone will upskill — become more capable, more autonomous, more strategic — and finally do the things they always wanted to do but couldn’t because of time and resources and headcount. One struggles to see how this constitutes a crisis.

The Flattening

Alex Karp is right that if AI displaces workers at scale, regulation will follow with a vengeance. But nobody’s following this thought to its conclusion.

The math will become obvious — and economics always wins — so companies that fired people will hire them back. But not into the same structures.

The org charts and the layers of middle management and the approval chains and the meetings about meetings — all of that existed because coordination was expensive and humans were bottlenecks. When every person operates at a hundred times their previous capacity you don’t need seven layers of management to ship a product. You need small autonomous teams with enormous leverage. The friction that chokes large organizations — the reason it takes four hundred people and eighteen months to deliver what a startup does in three months with five — becomes intolerable when the output ceiling is this high.

Companies will flatten. They’ll have to. The competitor with fifty empowered people producing at the level of five thousand will devour you while you’re still scheduling the pre-meeting for the meeting.

You need more generals, not fewer, because each one now commands an enormously larger operation. GDP doesn’t merely grow — it explodes. And instead of directing human genius toward optimizing the dimensions of an ad engineered to make people spend money they don’t have on things they don’t need, we work on things that actually matter. Larger projects. More meaningful projects. Projects that push the species forward.

The Real Concern

My real concern is different from everyone else’s. It’s not about employment. It’s about the softening of the species.

Hard people are needed to do what needs to be done — to stare into the abyss and fight on. The world throws curveballs at precisely the moment we’ve convinced ourselves the trajectory is smooth. What happens when electromagnetic pulses destroy the data centers on which civilization depends? What happens when the sun dies? The universe doesn’t care whether we’re ready for what it sends, and softness in the face of existential challenge is a luxury no species can afford indefinitely.

None of the anxious questions people ask today about employment will matter. People will work with extraordinary leverage and the economics will sort themselves out. But we must see the larger picture.

The Punchline

Here’s the punchline, and it’s almost precisely the opposite of what everyone seems to be saying: if you don’t have enough work for your people to oversee AI agents on, the problem isn’t that you have too many people. It’s that you’re not doing enough. You’re not being ambitious enough.

The world isn’t running out of work. It’s running out of people who can see how much work there is to do.

In the future we’ll do things because we want to, not because we have to. We’ve never run that experiment at scale. We’re about to.