• TheObviousSolution@lemm.ee
    link
    fedilink
    English
    arrow-up
    5
    arrow-down
    1
    ·
    22 days ago

    I think I’d choose a traditional CEO over an AI, AI can be quite the asshole when it begins hallucinating and can dig quite the ideological hole for itself, specially when the existence and nature of the company defines the goals and arguments it will adopt.

    Now HR, I’d love-hate seeing AI replace HR, knowing some of the assholes that make the rounds in it. Unfortunately, same problem with AI.

    The biggest problem with AI is that it only needs to fool us that it is making an intelligent claim.

    • deadbeef79000@lemmy.nz
      link
      fedilink
      English
      arrow-up
      8
      ·
      22 days ago

      That’s assuming the CEO isn’t already hallucinating.

      At least when an LLM hallucinates you can tell it and it won’t fire you.

      • TheObviousSolution@lemm.ee
        link
        fedilink
        English
        arrow-up
        4
        ·
        22 days ago

        It doesn’t have the power to do so. But it does have the power to shrug off your questions. Has an LLM ever shrugged off your questions?

        • deadbeef79000@lemmy.nz
          link
          fedilink
          English
          arrow-up
          4
          ·
          edit-2
          22 days ago

          Sort of, I had GitHub Copilot hallucinate an AWS Cloud formation template stanza.

          Asked it for the source it used for the stanza, which it then gave me the URL for.

          Told it that the crap it just gave me wasn’t on that page.

          It apologies and told me to RTFM.

          So, yeah, even super auto correct is a dick.