Also, to keep a theme going: This is just where my current thinking is going, not a slam-dunk. If you disagree with this, or even just think an important aspect is neglected, I would *love* to see a counterposition.
-
-
Replying to @catherineols
Some possible lines of counterargument: (1) why should we expect this pattern of compute beating human understanding to generalise to all kinds of future tasks/capabilities? Would be nice to see this spelled out. The examples given compelling, but are there any counterexamples?
2 replies 0 retweets 1 like -
Replying to @jesswhittles @catherineols
Stefan Schubert Retweeted michael_nielsen
Michael Nielsen disputed some of the examples:https://twitter.com/michael_nielsen/status/1106406616813821952 …
Stefan Schubert added,
michael_nielsen @michael_nielsenAn example is this argument that Deep Blue was just massive search. It wasn't: it used more than 8,000 hand-engineered features that incorporated a huge amount of chess knowledge. Contrast Sutton's essay on the left with the paper about Deep Blue on the right: pic.twitter.com/xyFQJH7LL9Show this thread1 reply 0 retweets 1 like -
I'm not sure that's disputed, so much as some of the examples are just flat-out wrong. But I would say that
The Deep Blue paper is full of stuff about the human chess knowledge they used, directly contradicting Sutton's essay. His other examples are only somewhat better2 replies 0 retweets 0 likes -
I just meant to say that you criticized some of the examples
1 reply 0 retweets 0 likes
Sure, got it!
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.