BrianChristian

• BrianChristian 20 Jan 2021 18:21 UTC

  1 point

  in reply to: BrianChristian’s comment on: Re­views of the book ‘The Align­ment Prob­lem’

  Ah, I almost forgot, here’s another posted to LessWrong itself:

  Review by Ben Hoskin (Larks)

• BrianChristian 20 Jan 2021 3:30 UTC

  11 points

  on: Re­views of the book ‘The Align­ment Prob­lem’

  If there’s interest, I’m happy to include a few here (including some from fellow LessWrong users):

  Review by The Wall Street Journal

  Review by Alison Gopnik

  Review by Joshua Gans

  Review by Owain Evans (Owain_Evans)

  Review by Alex Turner (TurnTrout)

  Review by Geoffrey Irving

  Review by Toby Ord

  Review by Martin Rees

  Reviews by Stuart Russell, Jaan Tallinn, and others

• BrianChristian 20 Jan 2021 3:08 UTC

  4 points

  on: No non­sense ver­sion of the “racial al­gorithm bias”

  >> False negative rate = (false negative)/​(actual positive)

  >> False positive rate = (false positive)/​(actual negative)

  Correct me if I’m mistaken, but isn’t it:

  False negative rate = (false negative)/​(false negative + actual positive)

  False positive rate = (false positive)/​(false positive + actual negative)