We think in
straight lines.
The world doesn't. Neither does AI. Our brains evolved to add. Reality multiplies. That one gap explains most of why the future keeps surprising us.
Feel it for yourself ↓Start at one.
Double it every second.
No tricks. Just times two, once a second. Watch how fast "small" becomes a number with no name.
Guess where
it lands.
Here are the first few steps of something doubling. Drag the handle to predict step 12. Then reveal the truth. Almost nobody guesses high enough.
Fold a sheet
of paper.
A sheet is a tenth of a millimetre thick. Each fold doubles it. You cannot really fold paper this many times, but the maths does not care. Keep folding.
This is the part
that matters.
Every instinct you just watched fail is the instinct we use to reason about AI.
The compute behind these models has roughly doubled every few months for years. Capability follows. And we keep meeting each jump with a straight-line reaction: impressive, but it won't do the next thing.
Then it does. On schedule. The curve was never going to warn us politely.
You don't have to predict the exact number. The doubling ticker never told you where it would stop, and it didn't matter. What mattered was the shape. Once you accept the shape, you stop being surprised, and you start planning for the part of the curve almost nobody is planning for.
We think in straight lines. The trick is learning not to.
Si