You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
There's nothing new added to this conversation on account of deep vs not deep. Basic questions about the limits of mining associations (vs reasoning) have been plumbed far earlier and far deeper by Rubin, Robins, @yudapearl, no?
The basic question IMHO is symbol-manipulation - do we need it or not? Two entirely different classes of problems. No real causal reasoning with out it, but people like Hinton and LeCun dismiss it, and even ridicule it (eg https://sites.google.com/site/krr2015/home/schedule …)
do you think non-human animals do "real" causal reasoning?
Tool use (ravens, primates etc) might be a good example.
If only "symbol manipulation" meant anything precise in algorithmic, statistical, or information-theoretic terms (contrast causal inference, for which @yudapearl et al provided precise distinctions). As it is, "symbol manipulation" is the AI/cogsci analog of intelligent design.
Double standard here: nobody complained about the clarity of symbol-manipulation when Hinton, Bengio, & LeCun argued against it in Nature 2015 (or when Rumelhart and McClelland did earlier in 1986). Or on a daily basis when they program with operations over variables!
The fundamental question is what kind specific kind of reasoning is necessary. Symbol manipulation is overly broad. If you said confined to 'constructive mathematics' then perhaps a point. However, you have not and it is an analog of intelligent design.
Douglas Hofstadter has a much more precise definition in his appeal to the mechanisms of analogical reasoning. However, demanding the need for 'symbolic variables' is even more nebulous than Chomsky's demand for the 'merge operator'.
They didn’t dismiss it; Hinton especially believes that the biological brain is a suitable inspiration of AI, and he explicitly acknowledges that brains don’t process symbols in such way. They made their stances and are working on solutions.
I believe we need logic and symbol manipulation, BUT not introduced as abstractions in the code, causing disconnection from rich environment's interaction. We need neural network architectures with emergent symbol and logic properties. Hint: inhibition, long time ignored
Yes - agree it's impossible to achieve AGI without symbolic reasoning for reasons that are too numerous for 140 chars
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
