It takes effort to be an informed citizen. Artificial intelligence tools offer an alluring shortcut — but they’re not without risk.

  • lennybird@lemmy.world
    link
    fedilink
    English
    arrow-up
    4
    arrow-down
    1
    ·
    15 hours ago

    Has anyone tested the prompts on these? I’m kind of wondering if truth having a leftist bias could help inform people against the right-wing propaganda.

    • taiyang@lemmy.world
      link
      fedilink
      arrow-up
      2
      arrow-down
      1
      ·
      10 hours ago

      I don’t think the truth is going to factor in, but my father has been outsourcing his thinking to Gemini lately and in that case it’s fairly straightforward: they have a profile on him as a moderate Democrat, so it suggests anyone with that tag. It likely isn’t going to fact check that tags, though.

      But back to truth: it will pull top search results which very much can be gamed, just like pre-LLM days. That means a false story pushed on a candidate is popular, it’ll default to that and could mislead a voter. That, and of course whatever weights the owner of that LLM sets (although, that is much more subtle than other commenters let on; a weight generally isn’t absolutely, it simply tips things so like, Democratic Socialists won’t magically get recommended MAGA but conservative independents will based on higher values set to conservative tags like immigration views).