• ShellMonkey@piefed.socdojo.com
    link
    fedilink
    English
    arrow-up
    27
    ·
    2 days ago

    Grok has shown the same ‘problem’ of being too determined to stick to facts despite their builder’s wishes.

    Even the bots are smarter than these clowns.

    • psivchaz@reddthat.com
      link
      fedilink
      arrow-up
      7
      ·
      1 day ago

      It’s actually something that gives me hope in the world. LLMs are predicting the words they use based on being fed tremendous amounts of data. The fact that they’re having trouble getting them to parrot lies is a good sign that the vast majority of data sources are actually agreeing on the truth. At least that’s how I view it.