Comments by "boz" (@BOZ_11) on "Consciousness is Not a Computation (Roger Penrose) | AI Podcast Clips" video.

I know Lex is totally bummed out by the idea that if computational power expanded 1 million times over what it is today, we still couldn't achieve consciousness synthetically. I don't think it's a bad thing that we never get to toy with something so devastating. We can't handle nuclear fission without a catastrophe and great loss of life every 30 years, never mind this monster.
5
@bhbluebird i mean yeah, but s.u. if consciousness is not a computational problem, and given that computers only solve computational problems our creations will always lack the I in AI, and therefore it's a massive misnomer.
2
@loicgrossetete9570 "AI as intelligent as an human is far away" - I think you misunderstand the size of the problem. It might well be outside our scope. Imagine attempting to teach algebra to a Labrador. Woof
1
@loicgrossetete9570 Intelligence has to be a feature of an actor which can turn to potentially any problem and at least begin working toward a solution (whether it finds it or not, not concerned about resolution). AI is coded such that parameters for a specific task must be set by the programmer (we're still OK here), the problem then is moving to another task means making substantial changes to the code and defining then adjusting the parameters for finding the solution to the new problem. If it needs constant input any time you're dealing with even a modified problem (not even an entirely new problem) then it isn't "intelligence" in any meaningful definition of the word. The machine doesn't understand what a "problem" is. It chisels away (removing wrong turns in its process) in a deterministic fashion faster than we could, but that's not intelligence.
1
30 is a baby. you won't know sh1t until 45
1