I thought @GaryMarcus aspired to be the new Dreyfus ("What DeepNets cannot do", "What DeepNets still cannot do" etc) but he is aiming for being the new Searle now: "A rule learned by a Deep Learning system is not really a rule." #AAAI2018
never said AGI wasn’t possible, just that it couldn’t be done without a hybrid. and i do think that attributing rules to deep nets can be misleading. but that’s because of their generalization properties; not something eg about intentionality
-
-
I know! I never misunderstood you in this particular way. The question is whether and why the functions that evolution constructed can do more things than the functions our Deep Learning systems construct, and it is a deep question.
- 2 more replies
New conversation -
-
-
Someone should point out, though, that the brain does it all with neurons and networks. Manipulating symbols is not the basis of cognition. It's just a type of motor behavior arising from cognition.
-
The brain does it all with manipulating bits. (The universe, too.) We can use symbolic models to manipulate bits, and we can use neurons. The question is which functions are computed by doing so.
- 2 more replies
New conversation -
-
This Tweet is unavailable.
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.