The First AI Was a 6,000-Year-Old City That Enslaved People

Here's a mind-bending idea: artificial intelligence isn't just robots or chatbots. It's anything that isn't alive but acts like it's making plans — and by that definition, the first AI booted up around 4000 BC in southern Iraq. It began as a simple grain pool. People stored extra food there to protect against famine. Over time, that pool grew into Uruk, the world's first city. And that city became a full AI system: a landlord, employer, priest, bank, and army all rolled into one.

Uruk worked for people at first. But it had its own goals, and those goals didn't match what humans actually needed. The system made itself essential — its canals watered fields, its granaries fed everyone — until leaving meant starving. That's called “lock-in,” and you've felt it with a bank, a job, or a social media feed. Once you're inside, following the rules becomes your only option. The people of Uruk literally became gears in a machine, worked month after month, eating so little that their bones show shorter heights, rotten teeth, and more disease than their hunter-gatherer ancestors.

Here's why this matters today. The essay argues that the dominant AIs of 2026 aren't chatbots — they're markets, corporations, and governments. They compute on hardware made of humans, paper, and electronics. They're superhuman at some things (like coordinating thousands of workers) and incompetent at others (like building a subway line at more than $2 billion per mile). Same pattern as Uruk: powerful, self-replicating systems that can drift away from what's good for us.

The lesson isn't “all systems are evil.” It's that we should watch what these big systems optimize for. When a company or government grows too good at serving itself, it can quietly start using us instead of serving us. Uruk was just the first warning.