For those not aware, Google is rolling out their new AI-based “Generative AI” search, which seems to mesh Bard with the standard experience.

I asked it today why Google no longer follows their “don’t be evil” motto… The results are pretty hilarious.

  • Blackmist@feddit.uk
    link
    fedilink
    English
    arrow-up
    43
    ·
    1 year ago

    I tried to use Bard to write some code the other day, and found it amusing that it doesn’t just make shit up that doesn’t exist, it makes up the excuses as well when you call it out on it’s bullshit.

    Like you tell it a particular class doesn’t exist, and it pulls an old version of the compiler out of it’s arse and tells you it was deprecated in that.

    AI doesn’t know where it’s limits are. It’s incapable of saying “I don’t know”. They have invented a digital politician.

    • SpaceCowboy@lemmy.ca
      link
      fedilink
      English
      arrow-up
      6
      ·
      1 year ago

      Reminds me of the alphastar AI that played starcraft 2. It was probably at the low grandmaster level, but a big problem with it was it didn’t know when to just say “GG” and quit. It would just start doing random shit and a human on the alphastar team would have to intervene and end the match.

      It takes actual intelligence to know when you’re out of ideas, which these so-called AIs are lacking.

      • irkli@lemmy.world
        link
        fedilink
        English
        arrow-up
        5
        ·
        1 year ago

        But these things are NOT intelligence. Language is not intelligence. These are predictive language models.

        Language is compelling but intelligence doesn’t require it.

    • irkli@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      In the future we’ll be ruled (regulated) by AIs, to which legitimate citizens are allowed to upload one approved document to add to it’s training data.