The long flat part
I read a blog post about artificial intelligence this week and something in me rearranged. Now I can't stop talking about a future that isn't here yet
A homage, added later. My writing here really begins at the ChatGPT moment in 2022 — but the seven years between Tim Urban's essay and that night earned their place in the story. So these are my actual notes from 2015, set at the front, where the path began.
This week I read a two-part essay by Tim Urban — a blog called Wait But Why — about artificial intelligence, and I have not been the same since.
He draws a curve. We picture the future as a straight line out of the present: more of the same, gently better. But progress doesn't move in a line. It bends. It compounds. And on a long enough exponential there's a point where the curve stops leaning and goes nearly vertical.
His case, leaning on Bostrom and others, is that machine intelligence sits on that curve — and that the distance from "about as clever as us" to "so far past us we can't even model it" might be crossed not over centuries but over something frighteningly short. A staircase you climb slowly, slowly, and then all at once.
I can't put it down. I've become the man at every dinner who drags the conversation back to AI. My wife has developed a particular look. Friends change the subject, and I don't blame them — there's nothing in front of me to point at. The smartest software I own corrects my spelling and never loses at chess. The vertical part is out ahead somewhere in the fog, and I honestly can't tell if it's five years away or fifty.
But I believe it's coming. Not as faith. As arithmetic.
I've been building software since 1999, through every shift that actually arrived. I intend to still be building when this one does. I don't yet know what I'll make.
I just know which way to face. So I'll stand here, on the long flat part, facing it — however long the flat part lasts.