The biggest mystery of all is how a network of neurons, which is a system of biological cells that transmits nerve impulses and exists in embodied living beings capable of movement through space, learning things, and feeling things, has anything in common with lines of code running algorithms inside a box containing a bundle of silicon and wires plugged into a wall somewhere that feels nothing, is immobile, and by definition cannot learn. (machines can not learn, even ones with neural networks. I did not write the dictionary or make up the definitions, blame society) Aside from n-e-u and r of course. Must be the use of gradient descent, or maybe weighted selection, or hidden layers. Definitely gotta be those hidden layers. So hard to find they are.