29
davide
1y

Nothing to say...

Comments
  • 2
    I wonder how you prepped it 😁
  • 5
    @UnicornPoo submissive AI is better than the alternative, I guess.
  • 12
    @iSwimInTheC
    "No, it should be 14."
    "And you should be in 3rd. Grade. Fuck you."
  • 3
    @c3r38r170 It would have been much better 😅
  • 5
    @netikras Chances are, there is no prepping needed. The correction feature is probably not guarded in any way because it would lead to cases where the AI is wrong and refuses to stand corrected.

    There is no algorithm for thruth. The AI is "just" a very advanced interface for querying a huge amount of crawled text.
  • 2
    @UnicornPoo They should really just cut it down to "Okay," + new "fact" - no statement about correctness nor any apologies.
  • 1
    @Oktokolo but what if anyone gets oFfEnDeD
  • 2
    @joewilliams007 Offended by what? It will still accept any correction - true or false.
  • 4
    This is exactly what I thought would happen when I first proposed this experiment.

    This makes the AI even larger minefield when it comes to "facts" or solutions, because it ultimately always assumes the user is correct...

    so if it's correct, you should not correct it, and if it's wrong, you're fucked, because you don't know if it's wrong, and if you do, then you didn't need to ask in the first place... even worse, if you correct it because of ignorance, you now gained the false sense of being right, because the AI suddenly agreed with you and apologized for it's initial statement.

    And this is why this was never a threat to anyone anywhere
  • 0
    Please don't make it stupid
    It's already pc to the point of self nullification like most of the manipulated heterosexual white men in America pretending to care about feminist values because they want some pussy
Add Comment