wiki:BigIdeas/AI

Ideas on AI

At the level of molecular biology there is no counting or numbers - only exact pattern-marching of unique structures or "symbols".

This means that all signalling is not digital, but symbolic, where each message is a unique symbol - a unique pattern. It uses a gradient-based instead of frequency-based regulation, without counting.

One of the implications is that the world at that level is pure-functional pattern-matching (enzimes and proteins) with asynchronous message-passing and gradient-based feedback loops.

The brain is an "analog" device, not digital. There is no numbers outside human mind.

Neurons, presumably, are propagating "weights", not bits, creating unique patterns in a mesh by firing specific sets of neurons. Like a cell, its language is of "structures" or unique symbols not numbers.

This, perhaps, was a great insight behind Lisps - symbolic computation by using the same list structure for code and data.

Not bad, yeah?

Last modified 4 years ago Last modified on Feb 16, 2017, 9:22:52 PM
Note: See TracWiki for help on using the wiki.