The only interesting question is "how?"
Here’s what I have been saying since 2001: “Neural networks have to be supplemented with the key components of symbol manipulation in order to capture the compositionality of language and thought.” Agree or disagree, @ylecun? At our debate you said “diagree”. Your stance now?https://twitter.com/ylecun/status/1065298686413078528 …
-
-
-
No, the interesting question is whether, and you have denied it, explicitly in print and conversation. Have you changed your mind?
- 5 more replies
New conversation -
-
-
Good for you that you thought that in 2001. I did, too. The problem is that people --- Turing award winners! --- had been trying to actually DO this for 50 years before the two of us thought of it. They got nowhere. In 2001, I didn't have any ideas how to do better. Do you, now?
-
the credit i deserve such that i do is for publicly laying out the case in great detail and standing ground in light of endless ad hominem attacks.
- 1 more reply
New conversation -
-
-
I respect you the most, sorry to interrupt but I think insisting on who said what is pointless from a scientific perspective. What’s important now is that everyone agrees on limitations of DL and want to improve on it. It’s an opportunity for cooperation. Let’s improve on it then
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Any recent papers with any approach other than ‘emergent substrate’?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The term "supplemented" seems to imply some extra functionality, like in $thinking_machine = nn + symbol_manipulator$. I wonder what that symbol_manipulator part could look like in a human brain?
- 1 more reply
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.