• @big_slap
    link
    -38 months ago

    it’s embarrassing how poorly it works when comparing it to chatgpt

    • @jacksilver
      link
      108 months ago

      I’m not sure this is unique to Google, isn’t this just an issue with how LLMs work?

      • @big_slap
        link
        -58 months ago

        I believe so. they need to have their generated answers pass through a basic knowledge check or something… I rarely get wacky answers when I use chatgpt, I wonder how they do it

        • El Barto
          link
          38 months ago

          Tell it how much is 30+30. Then tell it that it’s wrong, that the answer is 42. Then ask again how much 30+30 is.

          • @big_slap
            link
            58 months ago

            what was supposed to happen?

            • El Barto
              link
              18 months ago

              Oh! I stand corrected. I guess they fixed it.

              It should have told you 42.

              • @big_slap
                link
                28 months ago

                hahaha! it gets smarter every day, I guess.

    • @Larry
      link
      28 months ago

      First/ largest mover advantage, chatgpt gets more correction over a longer time from a larger userbase