The Machine That Builds the Machine

In 1956, Günther Anders made a prediction that shouldn’t have been possible. He said machines would stop waiting for us. That they would merge into a single system — he called it the “total machine” — and that system would begin to expand on its own terms, at its own pace, for its own reasons. Humans, he wrote, would become “co-historical”: still present, still watching, but no longer the ones deciding what happens next.

He was describing a factory floor. He could have been describing last Tuesday.

The first time I installed an AI coding agent, I gave it a task and sat back. It read my files, decided what was missing, wrote the code, ran it, hit an error, diagnosed the error, rewrote its own approach, and tried again. Nobody told it to do that second part. It corrected itself.

I sat there, genuinely overwhelmed. Not by what it could do. By what it couldn’t do without me. The system was fast, tireless, relentless. But it was only as good as the context I gave it. I was the creative brain. It was the statistical one. That gap — between what the machine can execute and what only a human can frame — is exactly what Anders spent his life trying to name.

He called it the Promethean gap. We can produce more than we can imagine. We build what we cannot feel. His sharpest formulation: “While ordinary utopians are unable to actually produce what they are able to visualize, we are unable to visualize what we are actually producing.”

The physicist Albert Bartlett put it differently: “The greatest shortcoming of the human race is our inability to understand the exponential function.” We think linearly. Machines compound. Anders went deeper: it’s not a cognitive failure. It’s a moral one. Our feelings have a ceiling. Our tools don’t.

He called this überschwellig — supraliminal. The effects of what we build are too large to feel, so we go blind. Not stupid. Blind. He called the condition “apocalyptic blindness” — not because the apocalypse is coming, but because our emotional apparatus cannot register the scale of what we’ve made.

“We can mourn a beloved dead person,” he wrote. “We can perhaps imagine ten dead.” But we can build systems that affect millions, and feel nothing at all.

Listen to how people in your company talk about AI. “We need to keep up.” “We can’t fall behind.” “We have to move faster.” Every sentence puts the human in the reactive position. The machine leads. The human follows. That’s the inversion Anders described, playing out in real time in your leadership meetings. Technology has become the subject of history. We have become its object.

Donna Haraway argued in her 1985 Cyborg Manifesto that the line between human and machine was already a fiction. Anders mourned the inversion. Haraway said: the merger already happened. The question is who gets to define its terms.

The answer, it turns out, is not us.

Every time you click “I’m not a robot” on a CAPTCHA, you are ritually demonstrating your humanity to a machine that gets to decide whether it believes you. Think about that for a moment. We’ve normalized this. We don’t even see the absurdity. That’s apocalyptic blindness in a checkbox.

And yet — here’s what people miss about Anders — he wasn’t a pessimist. He was a trainer. He believed the gap could be narrowed. Not by building better machines, and not by understanding AI more deeply. By training the human capacity to imagine, to feel, to choose in the face of what we’ve created. He called these “moral stretching exercises.” Treat your imagination like a muscle, he said. Because it is one. And it has atrophied.

The real question isn’t whether AI will keep building AI. It will. The question is whether we can keep imagining what that means. Not technically — morally. Can we feel the weight of systems we’ll never fully see? Can we make choices worthy of consequences we can’t yet picture? That’s not an engineering problem. That’s a human one. And it starts with stretching.

Anders knew this seventy years ago. Train the human. The machine will follow.

When was the last time your team made a strategic decision about AI — without using AI to make it?

---

Sources

• Günther Anders, Die Antiquiertheit des Menschen, Vol. 1: Über die Seele im Zeitalter der zweiten industriellen Revolution (C.H. Beck, 1956). English translation: The Obsolescence of the Human, trans. Christopher John Muller (University of Minnesota Press, 2025).
• Günther Anders, Burning Conscience: The Case of the Hiroshima Pilot (1961) — correspondence with Claude Eatherly on moral imagination.
• Donna Haraway, A Cyborg Manifesto: Science, Technology, and Socialist-Feminism in the Late Twentieth Century (1985), in Simians, Cyborgs, and Women: The Reinvention of Nature (Routledge, 1991).
• Albert A. Bartlett, “Arithmetic, Population and Energy” (1969) — lecture series on exponential growth and human cognition.