The arms of a galaxy wind in a spiral like the eye of a hurricane, a hurricane like the horn of a ram, a ram's horn like the shell of a snail. Not a resemblance. They are one curve, the logarithmic spiral, the only shape that can grow larger without ever changing its form. Jacob Bernoulli loved that curve past all reason, and had it carved on his grave beneath a single line: though changed, I rise again the same.

This is what Leonardo meant. Nature does not invent a new answer for every problem. It finds one that works and uses it again, wherever the same problem appears. Life is no exception: every living thing runs on a single code, and you and a jellyfish, who look nothing alike, are written in the same four letters. We call it DNA.

Modern AI rests on an assumption that intelligence is something you learn into existence, that if you pour enough of the world into a large enough machine, understanding will condense out of the data. It is the most expensive assumption ever made, and we believe it is wrong. Intelligence was never learned. It was inherited, written into structure by the same recipe that writes a body, shaped over hundreds of millions of years before its owner ever drew breath. The industry is trying to train its way to a mind. We think the mind is already written, and that the real work is not to manufacture intelligence but to read it.

No animal was trained into existence. A brain is built twice: once before birth, when the recipe hands it a structure it did nothing to earn, and again across a lifetime, as experience tunes that inherited structure. The first building is the scaffold, the second the polish, and the scaffold sets the limits of everything the polish can ever do. Today's machines invert the order: they begin with no structure at all and reconstruct the world from outside, by brute force, a blind search through oceans of data. It works, but it is the most expensive way imaginable to reach what nature arrives at for almost nothing.