The Outsourced Mind

Every tool we have ever built took something off the body. The plough took the back. The calculator took arithmetic. The map, and then the little blue dot, took our sense of where north is. We made the trade every time and mostly came out ahead, because what we handed over was labour, and what stayed behind — the judgment, the taste, the deciding — was still ours.

The machines being built now are different in kind. They are not coming for the labour. They are coming for the deciding.

You can watch it happen. People no longer ask the model only what to do; they ask it what to think. What to write, how to argue, which of two things is better, what a feeling means. The feed already decided what you would see today. Increasingly the model decides what you say about it. The loop that used to run through a person — notice, weigh, choose — now runs through a service, and the person signs off at the end without quite knowing what they agreed to.

Underneath that, the ground is worse. The attention economy spent fifteen years learning to capture a mind and hold it against its will, and it did this with human-made content and a budget. That budget is now zero. The slop is generated, endlessly, tuned in real time to whatever keeps you still. A whole generation is growing up marinating in it. Brainrot is the joke word for it, but the mechanism isn't a joke: a muscle you never use goes. Not dramatically. It just quietly stops being there when you reach for it.

I build these tools, which is the only reason I think I'm allowed to say this. I know exactly how good the offload feels — the change applies, the answer arrives clean and confident, you move on lighter. That feeling is not a bug in the product. It is the product. Nobody is going to ship you the version that makes you sit with the hard part longer. Convenience is the whole business, and convenience, past a certain point, is just capability quietly leaving the room.

Here is the part I think people get wrong about where this goes. The fear everyone rehearses is a superintelligence that turns on us — a mind so far past ours that it decides we're in the way. I don't lose much sleep over that one. The failure mode I actually see coming is duller and closer: a comfortable, mediocre intelligence that we lean on for everything, and a population that has offloaded so much of its own judgment that it can no longer tell when the machine is wrong. You don't need the machine to be malicious. You just need us to stop being able to check it. A species that can retrieve anything and hold nothing, that can generate everything and verify nothing, has not gotten smarter. It has gotten dependent, and mistaken the two.

The skill that survives all of this is the boring one: knowing what is actually worth deciding yourself, and keeping enough of your own edge to catch the answer when it's wrong. That isn't nostalgia for a harder time. In a world where everyone can summon a fluent, confident answer in a second, being one of the few who can tell a right answer from a right-sounding one is not a hobby. It's the last real leverage there is.