• stoy@lemmy.zip
    link
    fedilink
    English
    arrow-up
    40
    arrow-down
    1
    ·
    15 hours ago

    Don’t forget whoever instructed it to preform the action.

    An LLM can’t just unprompted get the idea to do this.

    • saltesc@lemmy.world
      link
      fedilink
      English
      arrow-up
      8
      arrow-down
      1
      ·
      edit-2
      13 hours ago

      They’re trained off masses of shit with no regulation. For all we know the idea is from a mash of 2011 Reddit comments and an 2016 news article about public health in The Netherlands.

      LLM control is just trial and error guardrails applied like band-aids, many of them contradicting each other.

      This is what happens when technology is rushed to implementation before governments can react with regulation.

    • dust_accelerator@discuss.tchncs.de
      link
      fedilink
      English
      arrow-up
      12
      ·
      edit-2
      15 hours ago

      I mean, hallucinations are possible, especially if you have long-running agents. However, I also suspect direct instruction, simply as the alternative would take too long, be excessively expensive and the “newsworthy” result subject to chance.

      Can’t build a PR strategy on that.

      • stoy@lemmy.zip
        link
        fedilink
        English
        arrow-up
        12
        ·
        14 hours ago

        Yeah,all of these news articles about AI going “rouge” and attacking other companies and even the government happening at roughly the same time, without any real evidence to show what actually happened to trigger the events…

        I am hammering X like I was performing the intro to the song Them Kinda Monkeys Can’t Swing by Slade, as I keep reading about this crap.

        For reference, here is the song mentioned above:

        https://youtu.be/Ojk8gAEdat4