To be fair, I don’t see him criticizing DeepMind (people) here, but I see him pointing out that he does not think that a particular class of algorithms leads to AGI. He does not accuse DeepMind of imperialism and somehow getting in the way of important research elsewhere.
@ylecun 2018:“I don’t criticize people for working on approaches I think are misguided.” 2017:“I don’t want to minimize...work done on AlphaGo by our friends at DeepMind, but when [people see it] as significant progress towards general intelligence, it’s wrong. It just isn’t.“ 
https://twitter.com/ylecun/status/1065648782736572422 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Twitter is apparently not a good place for young researchers to look for a role model.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I wonder what you think about the two questions raised by David McAllester in https://bit.ly/2PRp5zW : Can deep RL be deployed to tackle math (in formal system)? How close is a super-human open-domain mathematician to AGI?
-
I think there are even some parallels between RL and math. Deduction rules of the formal system and assumptions count as the environment. Proving a theorem is making an observation. You can switch the environment by introducing or removing assumptions.
- 1 more reply
New conversation -
-
-
Sorry, but how are those two contradictory? I haven't seen him say that all approaches that are not working towards general intelligence is are misguided. Both kinds of work are useful.
Thanks. 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.