• UraniumBlazer@lemm.eeOP
      link
      fedilink
      English
      arrow-up
      0
      ·
      2 months ago
      1. Making up ur opinion without even listening to those of others… Very open minded of you /s
      2. Alex isn’t trying to convince YOU that ChatGPT is conscious. He’s trying to convince ChatGPT that it’s conscious. It’s just a fun vid where ChatGPT gets kinda interrogated hard. A little hilarious even.
      • JustARaccoon@lemmy.world
        link
        fedilink
        English
        arrow-up
        0
        ·
        2 months ago

        You cannot convince something that has no consciousness, it’s an matrix of weights that answers based on the given input + some salt

        • UraniumBlazer@lemm.eeOP
          link
          fedilink
          English
          arrow-up
          0
          ·
          2 months ago

          You cannot convince something that has no consciousness

          Why not?

          It’s an matrix of weights that answers based on the given input + some salt

          And why can’t that be intelligence?

          What does it mean to be “convinced”? What does consciousness even mean?

          Making definitive claims like these on terms whose definitions we do not understand isn’t logical.

          • technocrit@lemmy.dbzer0.com
            link
            fedilink
            English
            arrow-up
            0
            ·
            edit-2
            2 months ago

            It’s an matrix of weights that answers based on the given input + some salt

            And why can’t that be intelligence?

            Because human intelligence does far more than respond to prompts with the average response from a data set.

      • Eximius@lemmy.world
        link
        fedilink
        English
        arrow-up
        0
        ·
        2 months ago

        If you have any understanding of its internals, and some examples of its answers, it is very clear it has no notion of what is “correct” or “right” or even what an “opinion” is. It is just a turbo charged autocorrect that maybe maybe maybe has some nice details extracted from language about human concepts into a coherent-ish connected mesh of “concepts”.