Comment on remember, if your gf isn't open source and running locally, you don't own her

doctorcrimson@lemmy.world ⁨8⁩ ⁨months⁩ ago

• FractalsInfinite@sh.itjust.works ⁨8⁩ ⁨months⁩ ago

  If all you need is a one sided conversation designed to make you feel better, LLM’s are great at concocting such “pep talks”. For some, that just might be enough to male it believable. The Turing test was cracked years ago, only now do we have access to things that can do that for free*.

  source

  • butterflyattack@lemmy.world ⁨8⁩ ⁨months⁩ ago

    A pretty early chatbot called Eliza simulated a non-directive psychotherapist. It kind of feels like they’ve improved hugely but not really changed much.

    source
  • doctorcrimson@lemmy.world ⁨8⁩ ⁨months⁩ ago

    Nah, bullshit, so far these LLM’s are as likely to insult or radicalize you as comfort you. That won’t ever be solved until AGI becomes commonplace, which won’t be for a long ass time. These products are failures at launch.

    source

    • FractalsInfinite@sh.itjust.works ⁨8⁩ ⁨months⁩ ago

      … Have you tried any of the recent ones. As it stands chatGPT and Gemini are both built with guardrails strong enough to require custom inputs to jailbreak, with techniques such as Reinforcement learning from Human Feedback uses to lobotomize misconduct out of the AI’s

      source

      • doctorcrimson@lemmy.world ⁨8⁩ ⁨months⁩ ago

        HaVE yOu trIED iT bEfOrE? fOr SIMplE tAskS it SaVEs Me A lOt oF timE AT wOrK

        JFC a skipping record plays right on queue whenever somebody speaks ill of the GPTs and LLMs.

        source

        • -> View More Comments
• Gabu@lemmy.world ⁨8⁩ ⁨months⁩ ago

  Is it really duping, though? The way I see it, most of these people are perfectly aware of what they’re doing.

  source

  • doctorcrimson@lemmy.world ⁨8⁩ ⁨months⁩ ago

    It’s a very nuanced situation, but the people being sold these products and buying them are expecting a sentient robot lover. They’re getting another shitty chatbot that inevitably fails to meet bare minimum companionship standards such as not berating you.

    There currently exists no ethical use of LLM AI. Your comment can be construed as defence of malicious people and actions.

    source

    • Cethin@lemmy.zip ⁨8⁩ ⁨months⁩ ago

      I’ve never met anyone who uses them, but I’m also not sure people actually think it’s sentient. I’m sure some do, but I’d assume the vast majority are just looking to have a conversation, and they don’t care if it’s with a person or a (pretty good) chat bot.

      Also, there is a way to use it ethically. As the post mentions, run it locally and know what you’re doing with it. I don’t see any issues if you’re aware of what it is, just as I don’t see any issue using auto-correct or any other technology. We don’t need to go full Butlerian (yet).

      source

      • Krauerking@lemy.lol ⁨8⁩ ⁨months⁩ ago

        Really? You don’t think anyone that uses these don’t think they are sentient?

        Sure it’s not like the people designing these are prone to make-believing the AI for sentient too, right?

        You are coming at this from your perspective which knows them to not be real. That’s not gonna be how the average moron thinks and there is more of them than you think. And they absolutely believe their is a tiny sentient brain somewhere in there that is alive. I’m all for people doing what makes them happy but also this is a loneliness confirming hole to get trapped in and absolutely opens doors to influence people through their imaginary friends that they think they can trust.

        source