Meanwhile, the defining characteristic of natural intelligence is its adaptability, its extreme generalization power
Show this thread
AI success largely consists of first narrowing down a problem until everything is well-controlled, then brute-forcing/hard-coding a solution
-
-
-
If you give a strong human LoL player a brand new character, they will be able to perform at an excellent level upon their very first game
Show this thread -
This, in my opinion, is a much better test of intelligence (generalization) than the ability to win the most games in a controlled setting
Show this thread
End of conversation
New conversation -
-
-
You still really, really need to check http://OpenCog.org
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
One caveat though. Today's deep learning AI fits that mold. But the much maligned Symbolic AI of 80s tried to solve big problems and failed.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yeah, that's the definition of coding, really
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
'Brute-forcing' seems unfair. What is the difference from humans listening and speaking for 2-3 years? We call it learning.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
'AI success' is an overloaded term? Success in industry may be to create user value != success in academia to pass Turing test?
-
I say this because I don't want to detract from the great work in all but the General AI context, but to learn from and celebrate them.
End of conversation
New conversation -
-
-
Certainly not how we should do it. Not how we do it at
@Numenta.https://youtu.be/-h-cz7yY-G8Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.