Comments by "Jim Luebke" (@jimluebke3869) on "Disagreement with Elon Musk about existential risks of AGI | Michael Littman and Lex Fridman" video.

  1. Wikipedia may have "values", but its mechanisms are not hardened enough against corruption / lies. Unfortunately, the mechanisms to harden it against corruption, when couple with its "values", themselves lead to corruption / lies. I don't trust Wikipedia editors to tell the truth about any subject that touches on their deeply cherished Politically Correct values.
    4
  2. "Smarter than themselves" -- define "smarter". An AI might define that to mean, "better at making paperclips" depending an idiosyncrasies of its initial priorities, training, or experiences. The failure mode of AI is going to be the Sorcerer's Apprentice problem. It picks one job, and does it too well.
    3
  3.  @fredfred7222  Eh, it's a little reductionist to say that all we do is survive and reproduce -- we do a lot of other things along the way, that other surviving / reproducing creatures don't manage to do. I would say that having the ability to turn an asteroid into a cathedral (given a bit of time to research the relevant tech) would make a universe that we existed all over, more interesting than the barren rocks there are out there now.
    1