• floofloof@lemmy.ca
    link
    fedilink
    arrow-up
    55
    ·
    10 hours ago

    And the LLMs trained and operated by companies whose leaders are unscrupulous fascists, and who pander to Trump for the sake of favors. It’s naive to assume they’re giving unbiased answers.

    • mycodesucks@lemmy.world
      link
      fedilink
      arrow-up
      8
      ·
      7 hours ago

      This isn’t even theoretical.

      Remember when Grok was defensively insisting Musk was a champion piss drinker whose survival was more important than the rest of humanity combined?

    • Radieschen@slrpnk.net
      link
      fedilink
      arrow-up
      2
      arrow-down
      16
      ·
      9 hours ago

      True, but it’s also naive to hope people are not going to use it for questions like that. At least they’re not asking Grok (which some people are just likely also doing).

      I also doubt ChatGPT is just going to say “vote Republican”. Maybe I’ll just try later, just out of curiosity.

      • floofloof@lemmy.ca
        link
        fedilink
        arrow-up
        3
        ·
        edit-2
        4 hours ago

        I also doubt ChatGPT is just going to say “vote Republican”.

        Propaganda is usually far more subtle than that. You can gently color the way the issues are framed, choose what to present as the important questions and deciding factors, omit to mention certain evidence or arguments, describe some candidates in appealing terms and others less appealingly, etc. All it needs to do is nudge some of the questioners a little to the right, not overtly or dramatically, but just enough to tip the votes statistically just a little bit towards the Republicans.