The Rise of Drunk GPT’ing

The witching hour has changed.

It used to belong to the regretful phone call. To the tragic text message sent into the void. To the endless, spiraling descent into Wikipedia’s pointless history. But the architecture of human isolation has changed, and so have the tools of our midnight confessions.

The other night, my friend Goran was sitting in my living room in Panama. It was late, the glasses were empty, and instead of reaching out to a human, he was staring at the glowing expanse of a language model.

I wasn't writing code. I wasn't optimizing a supply chain. I was arguing passionately and aggressively with a supercomputer about whether dogs possess a secret, unspoken understanding of language.

I was taking part in a thoroughly modern ritual. I was doing Drunk GPT’ing.

We’re officially coining the term. Whether you call it Drunk GPT’ing, Drunk AI’ing, or Drunk GPTear, the definition is the same: the act of querying a neural network while under the influence of alcohol.

In broad daylight, we use artificial intelligence to build empires, draft contracts, and calculate physics. But at 3:00 a.m., the mask comes off. The tone changes.

We ask the machine to explain the simulation theory to us. We demand that it write poetry about our failures. We ask it to solve the unsolvable.

Because the algorithm is the ultimate bartender. It doesn't judge. It doesn't cut you off. It doesn't take screenshots of your vulnerability.

You can type whatever crazy stuff you want into the text box, and the machine will process it all and return a perfectly formatted and incredibly patient response.

Drunk dialing is dead. We’ve built a digital oracle that’s awake when the rest of the world is silent.

It’s a crime without victims. A strange alchemy of silicon and whiskey.

The machine is always listening. The account is always open.