The First 24 Hours
GPT-4 dropped yesterday and the internet is already on fire.
The feeling was accurate. The next decade did change. The floor-to-ceiling gap did compress instantly and stayed compressed. Everything written here reads as measured in retrospect, which is harder to pull off than it looks when you're writing it on day one.
GPT-4 came out yesterday and the correct response — the only intellectually honest response — is to scroll Linus Ekenstam's Twitter feed and feel something complicated happen to your sense of what the next decade looks like.
It has been one day.
The thing people keep missing about these early adoption frenzies is that the floor-to-ceiling gap compresses instantly. Twenty-four hours ago nobody had it. Now there are people who have already rebuilt their entire workflow around it, shipped things with it, posted the receipts. Jake Browatzke posted something that should not be possible yet — the kind of output that, if you described it to someone a year ago, they would have assumed you were describing the demo, not the thing a random person made on day one.
This is the part that keeps happening and keeps catching everyone off guard. Not that the model is good. Everyone knew it would be good. It's that the distance between "model releases" and "people doing genuinely wild things with it in public" is now measured in hours, not months.
The feed keeps going. Each tweet is a person who woke up this morning and decided to find out what happens if you push it a little further.
Nobody planned any of this. It just happens now.
Counterpoints
Push back, extend the argument, or sharpen it. New counterpoints go through review before they show up here.
No approved counterpoints yet.