When German journalist Martin Bernklautyped his name and location into Microsoft’s Copilot to see how his articles would be picked up by the chatbot, the answers horrified him. Copilot’s results asserted that Bernklau was an escapee from a psychiatric institution, a convicted child abuser, and a conman preying on widowers. For years, Bernklau had served as a courts reporter and the AI chatbot had falsely blamed him for the crimes whose trials he had covered.

The accusations against Bernklau weren’t true, of course, and are examples of generative AI’s “hallucinations.” These are inaccurate or nonsensical responses to a prompt provided by the user, and they’re alarmingly common. Anyone attempting to use AI should always proceed with great caution, because information from such systems needs validation and verification by humans before it can be trusted.

But why did Copilot hallucinate these terrible and false accusations?

  • finitebanjo@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    2 months ago

    lmao yeh bro such a hard riddle totally

    I concede. AI has a superintelligient brain and I’m just so jealous. You have permission to whip me into submission.

    • WldFyre@lemm.ee
      link
      fedilink
      English
      arrow-up
      1
      arrow-down
      1
      ·
      2 months ago

      What on Earth is this in response to?? Did I say it was a hard riddle?

      I concede. AI has a superintelligient brain and I’m just so jealous.

      Point to any part of my comment that implied any of this.

      I only gave more info on how LLMs work since what you were describing were Marcov chains. I wasn’t saying you were wrong with the thrust of your comment, just the details on how they work. If they were exactly as effective as Marcov chains we wouldn’t be having these discussions, that’s why they can be misused.

      Feel free to discuss the actual words I’m using instead of this LLM word salad.

      • finitebanjo@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        2 months ago

        Well clearly the AI couldn’t solve the riddle because even humans find it soooooo haaaard, daddy.

        • WldFyre@lemm.ee
          link
          fedilink
          English
          arrow-up
          1
          arrow-down
          1
          ·
          2 months ago

          Are you in high school? You’re making up things I never said and putting a sexual element on your responses for no reason.

          Stay in school and learn how to have discussions before arguing about language and technology lol

          Goodnight.

          • finitebanjo@lemmy.world
            link
            fedilink
            English
            arrow-up
            1
            ·
            2 months ago

            I’m just sick of you morons extending these comment threads out 5, 8, 20 fucking replies in defense of one of the shittiest things mankind has ever invented.

            • WldFyre@lemm.ee
              link
              fedilink
              English
              arrow-up
              1
              arrow-down
              1
              ·
              2 months ago

              Yeah, Marcov Chains are truly the worst thing our species has produced!

              Glad you have your priorities straight. It’s been fun talking to a chatbot instead of having a discussion like normal people. You can respond to this comment with whatever responses you want, just know that according to the things I’ve pretended you said, I’ve won the argument.