So you’re saying, I should just throw another TPU at it?! LETS SHIP IT! ;)
I don't understand "deep learning maximalism" (the idea that DL is the only piece of ML worth remembering & investigating). A view often held by the same absolute clowns who think all we need to get to human-level intelligence is a bigger DL model.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
b-but tHe ArChItEcTuRe Of AgI iS pRoBaBlY a TrAnSfOrMeR...
-
Damn, this tweet cracked me up
Maybe cause I read all your earlier tweet in todays argument
End of conversation
New conversation -
-
-
I like
@pmddomingos’s view in his “Master Algorithm” book that having multiple competing AI approaches that learn from each other will get us there: neural networks, graphs, symbolic, causal, clustering, etc. And new stuff that’s totally different. -
This will only serve to make less narrow, narrow AI. AGI is actually a highly specific learning process and cannot be captured by a mash of output focussed techniques.
- Show replies
New conversation -
-
-
look, there are two ways of solving a problem in tech : throwing as many buzzwords as possible to it, or choosing one and getting the buzzword very big, and then throwing it at the problem.
-
What inputs are required to grow that buzzword in the second process? I figure I could create some sort of middleware startup and glom me some sweet investor loot.
End of conversation
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
When someone improves the perceptron to be fundamentally another step closer to a real neuron, that's when the next big leap happens in DL... We need to go back to the beginning to take the next leap forward...!
-
But have we fully understood real neurons yet?
- Show replies
New conversation -
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.