Heady, Esoteric

Message:

I have coded on that machine in binary. But to crunch numbers not create a prediction model of a process.

The guy is teaching the computer to do a simple visual task. He doesn't force (in code) the machine to do the task of flipping numbers in a sequence. He sets up a prediction framework that in the end will reverse the numbers. Kinda neat but very specific. How many THINGS (weighted processes) must true AI discover and hold as interrelated models?

There are some difficult CS terms used here. I understand the guess a number and then see the result, watching parameters in an equation converge or diverge. I've done that (in BASIC). Knowing only five of six simultaneous equations and brute-force guessing the sixth. That's what I see in this.

Yet idea of setting up a weighting grid for a seven digit number and then correcting it based on how lucky the machine guess? That's more arcane.

Moving to understanding a dialect of Navajo or making an animated film seems more than incremental guess. Maybe I can't conceptualize the code at scale. I am not trained in CS nor understand concepts of AI.

Are human brains (babies) doing the same thing? Guessing what happens and filing the result. One-year-old dropping things, then throwing things, then dropping heavy vs light, on soft vs hard floor. Makes a mess often.

=====

AI will change work. That revolution will aid the few. Don't expect free-cash to result and be divided equally among society. This is not Irrigation or Industrialization. Which gave more to the many, plus needed people doing tasks. AI specifically does not need people. Maybe that's the first process model it calculates, "How can we eliminate the need for people doing X-Y-Z?" Did we? Try again until we do so, completely. It's less a tool of the wise and more a replacement of 'umans.

Yours,