does anyone really think the class "humans" can have a coherent anything, much less coherent extrapolated volition
-
-
Replying to @literalbanana
Tacit assumption of rationalism is that humans are rational in their essence and only contingently irrational due to
1 reply 0 retweets 6 likes -
Replying to @Meaningness @literalbanana
accident of running on lousy hardware. We are shadows of our Platonic Ideal selves, who have tidy mathematical definitions.
1 reply 0 retweets 8 likes -
Replying to @Meaningness @literalbanana
Benevolent AI could easily see which Ideal we each correspond to, and replace the corrupt wetware with that.
2 replies 0 retweets 7 likes -
Replying to @Meaningness
not multiple incompatible ones for each person, much less different people?
2 replies 0 retweets 1 like -
Replying to @literalbanana
And they think Aumann’s Agreement Theorem proves that we’d all be identical if only it weren’t for wetware limitations.
2 replies 0 retweets 2 likes -
Replying to @Meaningness @literalbanana
Or, put another way, there is a unique optimal response to any input, and after the Singularity, we will do that.
3 replies 1 retweet 1 like -
Replying to @Meaningness @literalbanana
Only given identical value functions.
1 reply 0 retweets 2 likes -
Replying to @GrumplessGrinch @literalbanana
Yes… diversity of which does pose a problem for “coherent extrapolated volition”…
2 replies 0 retweets 1 like -
Have you guys read the original doc? It tries to deal with this, eg p35 https://intelligence.org/files/CEV.pdf
2 replies 0 retweets 2 likes
The proposal is not "each human has the same ideal utility function X, calculate X"
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.