• NeptuneOrbit@lemmy.world
    link
    fedilink
    arrow-up
    190
    ·
    7 months ago

    Well it had the most up votes! So clearly it was the most generally useful and true response in the whole thread!

    • Norgur@fedia.io
      link
      fedilink
      arrow-up
      66
      arrow-down
      6
      ·
      7 months ago

      Because on Reddit (and here on Lemmy) people use up and downvotes exclusively to rate the quality of a post, not as a tool to show disapproval without having to be able to actually articulate why. I like when social interactions go exactly as intended - like in this case - and don’t devolve into two people arguing with silent mobs behind them.

      /s for everyone who’s as blind to sarcasm as this shitty AI from Google.

      • NeptuneOrbit@lemmy.world
        link
        fedilink
        arrow-up
        45
        ·
        7 months ago

        Humor IS a quality though. Human voters are able, in aggregate, able to award certain types of humor. Which an LLM is not able to. Which gets recycled with no context, as fact.

        The internet would be a vastly different place if sarcasm and in jokes were not regarded as a type of “quality” content.

        • pory@lemmy.world
          link
          fedilink
          English
          arrow-up
          4
          ·
          7 months ago

          Well, there’s an oddly existential argument to be made that “funny” AI answers like this or adding a pot of Elmer’s Glue to pizza sauce to get the cheese to stick are valued. Simply because those are the posts I’ve seen from Google AI, and I’ve never touched the feature myself.

          By letting a language “speaker” learn from Reddit and forums, we created an approximation of “that guy who thinks he’s a comedian” because that guy is always there and always drowns in upvotes. Clearly, he’s a valuable part of the discourse!