• Sludgehammer@lemmy.world
    link
    fedilink
    English
    arrow-up
    29
    arrow-down
    1
    ·
    edit-2
    12 days ago

    X declined to comment, but its AI chatbot, Grok, indicated the clip had been restricted under the Online Safety Act due to violent content.

    Oh my god… it doesn’t work that way!

    The chatbot doesn’t know who restricted the video or how it was restricted. The chatbot doesn’t know anything, it just calculated that based on its training data “Yes the video was restricted due to the Online Safety Act” was a likely answer and then it spit that out.

    • kibiz0r@midwest.social
      link
      fedilink
      English
      arrow-up
      6
      ·
      edit-2
      12 days ago

      Not reporting on it kinda lets X have it both ways though. They can curate it to be a spokesperson and default to spreading a certain message to users, but if the media reports on it spreading that message it’s suddenly out of context because it’s just a fancy autocomplete.

    • Flax@feddit.uk
      link
      fedilink
      English
      arrow-up
      4
      ·
      12 days ago

      I remember ChatGPT telling my dad that his image will be generated any moment now when the app clearly glitched out and didn’t give him his 20th Ghibli image of my mum.