If we ever build a machine that can, on it's own, search for new questions in mathematics, deduce new conjectures, and be able to prove them, then we can be sure that Norwig was wrong, and that we are doomed.
Until that point AGI is a meme.
Do you agree with me? Arguments for/against such machine. Discuss.
>>7930478
>Do you agree with me?
Agreed. It is impossible to replace man.
>>7930478
No? Math is meaningless drivel that doesn't ensure any kind of survival fitness.
Fucking autists I swear to god.
If a computer can expound upon data sets we'd have to have a definitive, programmable, implementable and encompassing understanding of number theory.
>>7930478
>Norwig was wrong, and that we are doomed
These two things are orthogonal. A weak AI could destroy the human race just as easily as a virus could destroy the human race.
But to answer the question you actually intended to ask, yes, strong AI is going to be a reality.
>>7930478
That has already been done; see, for instance, Lenat's Automated Mathematician, or any automated theorem searcher.
It's simply irrelevant. The problem is that they all, eventually, run out of new interesting questions, and their self-improvement hits a wall. They either reach a point where their mechanical abstract reasoning runs into the limits of what logical leaps and generalizations it can make, or else their heuristics and pattern-matching reach a point where they can no longer couple them to better reasoning well enough. And of course, they all require supervision and hand-tuning.
If we could simply keep discovering new theorems and extending our capacities infinitely, it would be a general AI. That's what general AI means. So yes, if we build an AGI, we can definitely build an AGI.
And what does "AGI is a meme" even mean? That we still require major conceptual breakthroughs and qualitatively different approaches to create it, and thus can't build it currently? Nobody seriously involved in AI research disputes that. That it's simply an idea, in the original sense of meme? Then what you're really asking would seem to be "Until the point at which we build a AGI (Which a broad unrestricted mathematical AI would indeed be), AGI is simply an idea that does not exist?" In which case, duh.
>>7930478
Richard Feynman Computer Heuristics Lecture
https://www.youtube.com/watch?v=EKWGGDXe5MA
>>7930715
>A weak AI could destroy the human race
Christ you fucking children. Go read a fucking book.
How about you start with Norwigs book for example.
Fucking dumb kids, get the fuck out.
>>7930670
>Math is meaningless
>He types on a computer, to post on a Tuvan throat singing appreciation forum.
>>7931822
Maybe when the pleb refers to math he might refer to Interuniversal Teichmüller. For, at this point, it is drivel.
Goddamn fucking norvigposters
as far as I know, norvig has never been wrong his whole life. SO WE ARE SAFE
>>7931825
PETER NORVIG > SHINICHI MOCHIZUKI > PERELMAN > ANDREW WILES > RICHARD STALLMAN > NEIL DEGRASSE TYRON > DEMIS HASSABIS
