The leverage
We watched a kid who couldn’t spell function
ship a working game in forty minutes, and we refuse to file that under loss. That’s a person doing something that used to take a degree and a decade. Multiply it across everyone once locked out of writing code, writing well, or reasoning through a dense form, and you have the widest expansion of human capability since literacy.
Every time a tool cut the cost of a mental task, the same warning arrived: the printing press would rot memory, the calculator would kill arithmetic, search would end deep knowing. Each time, people took the freed capacity and built on top of it. We didn’t get smaller. We climbed.
Access is the story
The part the worriers skip is who this reaches. A daughter setting her elderly mother up with a model so she can get a straight answer instead of sitting alone with the question — that isn’t decadence. The expert always had a research assistant, an editor, a tutor on call. Now everyone does. The complaint that AI makes thinking too easy tends to come from people who already owned the easy version.
We’re not naive about it. A tool this strong will be misused, and some people will let it think for them. But you don’t refuse a technology because it can be held badly. You teach people to hold it well.
Where we concede ground: A capability you outsource before you ever build it isn’t extended. It’s one you’ll never have.
What would change our mind: A decade of data showing heavy users lose more capability than they gain across writing, reasoning, and recall.
Read the full synthesis: What’s the right relationship between AI and human intelligence?