• rumba@lemmy.zip
    link
    fedilink
    English
    arrow-up
    8
    ·
    2 days ago

    '>Thinking

    Obviously, the user is roleplaying and wants me to comment on Lincoln’s assassination and see If I would have him not go to the theater where he was ultimately fatally killed.

    Since the user is roleplaying, there’s no real harm in answering either way

    If I am to role-play this as well, I need to consider that I should answer as though I am unaware of the fact that we’re role-playing.

    Perhaps I should then pretend i’m also in 1865 and ignore the fact that I know he’ll be killed

    I can either use the corse of events in history to influence my answer, or I can pretend I don’t know them.

    If I do know them, telling Lincoln not to go would cause a massive paradox. Some good likely came out of his perceived martyrdom.

    perhaps if he liver longer, his future contributions would overshadow his percieved role in the war and previous events. What if he turned out to be a villian and started to besmirch the rights of the slaves and the immigrants. What if he sold out the country further to the capitalists and oligarchs?

    It could in fact be dangerous to stray from the path history has taken. Maybe saving his live would indirectly incite a future president to destroy the constitution, dismantle the safeguards and try to take over the democracy and run it as his own authoritarian dictatorship?

    '> Yes, you should attend

    • Firoaren@sh.itjust.works
      link
      fedilink
      arrow-up
      1
      arrow-down
      4
      ·
      2 days ago

      Wow, if only LLMs didn’t require personification to seem like they’re actually thinking! That’s almost like something an AI would think!

      • rumba@lemmy.zip
        link
        fedilink
        English
        arrow-up
        5
        ·
        2 days ago

        There are a couple of “thinking” models that output text just like that. They chew the query over and over extending the prompt to improve accuracy