Is hard AI really impossible?
dunno but there's a lot of promising work being done on neural networking that surpasses human cognition so it doesn't really matter.
>>7713696
any examples?
>>7713891
He's going to bring up DEEPMIND beating the 600th best Go player at Go. Wow.
>>7713891
computers can beat the best human players in many fields, and are even able, in certain instances, to diagnose better than human doctors (?)
>>7713639
It hasn't been proven to be impossible.
I think that real, sentient AI will have to come from ALife field rather than the "deep learning" field.
If we can simulate 900 Million years of evolution, we can (increase probability of sentient, complex life emerging from it).
The deep learning isn't concerned with sentience, only problem solving. So things like identifying objects, playing games, etc... Those are goal-based optimisation problems which are easy enough to define for computers but doesn't mean jackshit for consciousness.
>>7713907
Fields like Donkey Kong 64, Checkers, Chess, Pac Man, MGS2, and Dreidel, Dreidel, Dreidel.
Better at diagnosing... Heart attacks. And it's not computers, it's an algorithm that fits on a sheet of paper, a fucking checklist. So the checklist diagnosed Heart Attack better than doctors. Wow! AI rebel-lution!
>>7713923
>So things like identifying objects, playing games, etc... Those are goal-based optimisation problems which are easy enough to define for computers but doesn't mean jackshit for consciousness.
This so much - all this "this is big news for AI!" hype around Go by Yudkowsky and these other American mouthbreathers doesn't mean anything - the computer just got better at optimising the goal of the game, not at becoming an artificial intelligence