[ExI] DeepMind wins Game1 in Go championship Match
sparge at gmail.com
Thu Mar 10 13:21:27 UTC 2016
On Wed, Mar 9, 2016 at 3:45 PM, BillK <pharos at gmail.com> wrote:
> DeepMind *did* teach itself to play Go better.
No, its programmers "taught" it learn how to play better. It didn't say
"Hey, you know what would help me play better? Neural networks that learn
from zillions of simulated games which kinds of moves work. Hang on...OK
that's done." AlphaGo is a Go idiot savant. It's a special-purpose human
construction that can do one thing well, but can't do anything else, and
doesn't really understand what it's doing.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the extropy-chat