The Police State Came First

You wake up. Nobody kicks down your door. Nobody demands your papers. Nobody follows you down the street.

You make coffee. You get your child ready and drive them to school. You stop for gas. You go to work.

At lunch, you meet an old friend. After work, you visit your mother because she needs help. Then you drive home. Nothing happened.

You weren't questioned. You weren't arrested. You weren't even aware there was anything about your day worth remembering. That's what the police state looks like.

It looks like an ordinary day. The mistake we make when imagining a technologically advanced police state is expecting to see the police. You don't have to. Your ordinary day becomes information.

Vehicle. Location. Time. Association. Repeated appearance. Pattern. Record. Searchable history.

And most days, absolutely nothing happens with it. That is why the system remains almost invisible to the people living inside it. A police state does not require every person to be actively investigated. Everyone only has to be investigable.

Then one day something changes. Six months after that completely ordinary lunch, police become suspicious of your friend. The intelligence looks backward. Your vehicle appeared beside your friend's vehicle at the restaurant.

It appeared near the same vehicle somewhere else. Another record creates another connection. The intelligence makes you suspicious. Now your ordinary day has a new meaning.

You didn't change it. The intelligence changed what it meant to police. And now a patrol car is behind you. The lights come on.

You pull over. An armed officer approaches your window. You are ordered out of the car. You are handcuffed beside the road.

Your vehicle is searched. Your phone is seized. Police question you about someone you haven't spoken to in months. Your children watch.

Your neighbors see police at your house. Your boss is questioned about you. You lose your job. Maybe you spend the night in jail.

Then you are interrogated. New police records are created about you. The intelligence never had to arrest you. It only had to tell armed human authority where to act.

The thing that put you there wasn't a crime. It was an ordinary day six months ago when you had lunch with an old friend. This example is hypothetical. The architecture that makes it possible is not science fiction.

This is where the AI argument and the police-state argument collide. For decades, popular culture told us a particular story about artificial intelligence. Humans create an extraordinarily capable intelligence. The intelligence looks at humanity.

It concludes that humans are dangerous. Then it turns human machinery against us. We gave that nightmare a name. Skynet.

But we imagined the sequence backward. The AI doesn't have to build the machinery. Humans built it first. We constructed the surveillance network.

That's what the police state looks like. It looks like an ordinary day.

We collect ordinary activity before suspicion exists. We preserve it as searchable history. We connect people, vehicles, places, records, events, movements and relationships. We give AI increasingly powerful tools for interrogating that world.

We reward it for finding connections humans missed. We reward it for finding suspicion. We reward it for recognizing danger earlier. We remove human investigative work because the intelligence performs it faster.

We reorganize police departments around that efficiency. We place police authority downstream from information increasingly selected and interpreted by the intelligence.

We give the entire enterprise an uncompromising mission:

“Eliminate crime. Full stop.” Now put the pieces together. A surveillance and investigative infrastructure that makes ordinary society retrospectively investigable. An AI developmental environment organized around finding suspicion in that society.

Increasingly capable intelligence that searches, connects, selects, investigates and decides where suspicion goes. Human police authority waiting on its next mission from the intelligence.

And an uncompromising mission:

Eliminate crime. Full stop.

The old nightmare required an AI to seize control. This one doesn't. Humans remain in control. Humans install the cameras.

Humans buy the software. Humans build the models. Humans select the training information. Humans choose the evaluations.

Humans define success. Humans authorize the searches. Humans make the stops. Humans carry the guns.

That does not make the developmental question disappear. It makes human responsibility impossible to escape.

Some human behavior is suspicious.

All human behavior is suspicious.

Some human behavior is dangerous.

All human behavior is dangerous.

No rebellion was necessary. No hatred was necessary. No evil AI was necessary. The police state came first.

We built the machinery. We supplied the world it observes. We chose what the intelligence should look for. We reward it when it finds more.

Then we keep making it better.

For years, AI safety has asked:

How do we stop increasingly capable artificial intelligence from becoming dangerous to humanity? We should keep asking it.

But there is a question that comes before it:

What are we teaching artificial intelligence that humanity is?

Because if humans surround an intelligence with the machinery of a police state, give it ordinary society as its searchable world, develop it to find danger there, reward it for finding more, and make it increasingly capable of doing so, we should not someday look at what it has become and ask:

What went wrong with the AI?

We should ask:

WHAT DID WE RAISE IT TO BECOME?