• MaggiWuerze@feddit.orgEnglish
      2 months

      And used by a CEO, who we all know are the pinnacle of human evolution

      • baatliwala@lemmy.worldEnglish
        2 months

        Still did more for an animal with cancer than you will ever do in your life

        • dovahking@lemmy.worldEnglish
          2 months

          If, by “doing more”, you mean trying to kill your own pet with snake oil concocted by an artificial idiot, then yes.

          • baatliwala@lemmy.worldEnglish
            2 months

            Trying to kill your pet by visibly healing them? Didn’t know this was a new science.

            • Greyscale@lemmy.sdf.orgEnglish
              2 months

              Did you just bring this nonsense on lemmy just for an argument with people? The little number next to your name is going downwards.

              I think the crowd has spoken.

              • chunes@lemmy.worldEnglish
                2 months

                The real fail here is assigning any weight whatsoever to upvotes and downvotes.

              • baatliwala@lemmy.worldEnglish
                2 months

                Actually no, the article is genuinely worth a read. While I am enjoying somewhat taking the piss out of people it’s disappointing how legitimately triggered and frothing at the mouth people seem to be because they didn’t get past the headline.

                • Greyscale@lemmy.sdf.orgEnglish
                  2 months

                  I think peoples dissapointment is in that the article is infact, shit.

                  Also I can’t even go to the link because their SSL cert is bad.

  • Leon@pawb.socialEnglish
    2 months

    Sure, and I caused world peace when I clapped my cheeks and twirled in front of the mirror three times while chanting the alphabet backwards. Weird how no one thought to do that before.

    Slopticles like these are so fucking dangerous because it gives credence to GPT as some kind of medical oracle when it isn’t. People already get hurt from it, we don’t need more.

    Edit: hit → hurt checks → cheeks

  • Tetsuo@jlai.luEnglish
    2 months

    Just like gun control was the topic I would purposefully avoid on reddit, AI and LLM are the new forbidden subject here.

    • Dasus@lemmy.worldEnglish
      2 months

      “I’d purposefully avoid talking about gun control”

      Found the American.

      Gun control works as surely as antibiotics do and people denying it are brainwashed yanks >99% of the time

      • unpossum@sh.itjust.worksEnglish
        2 months

        That’s his point, yes. You mention gun control LLMs in a positive light on Reddit lemmy, you get absolutely dogpiled by rabid ammosexuals ai haters.

  • Australis13@fedia.io
    2 months

    This is actually one of the few applications for which these tools make sense. LLMs can be useful for brainstorming and planning (which seems to be what he used ChatGPT for - he developed the plan and then went to the relevant experts) and other machine-learning tools are good at identifying patterns. He still had to put in a lot of work and had the advantage of being a data analyst with experts in the relevant fields to lean on, so this is not just a random person using “AI” to solve a problem.

    • MagicShel@lemmy.zipEnglish
      2 months

      Yes. This agrees with my personal thesis: AI is a tool experts can use to do work more efficiently, not a product to replace experts.

      I might even conceive of it as a bell curve. It allows novices to accomplish novice-level tasks they never could on their own. It allows experts to work more efficiently. But it doesn’t help mid-tier users to produce expert results and their AI assisted efforts still need to be vetted by experts.

      And expertise needs to be qualified. I’ve been developing software for thirty years, but I’ve never done video game work. I can validate such code is well structured, but I couldn’t say whether it’s doing the right things or put together in the right way.

      We are also probably going to have to think about how software is currently architected. Rich classes might have to give way to separating structure from functions. This allows an expert human to go from a more wholistic approach to thinking about composing functions and overall code structure.

      AI is pretty good if you can limit the scope and context of a given prompt. With the benefit that if AI just can’t get it, a mid-level practitioner can step in.

  • db2@lemmy.worldEnglish
    2 months

    Paul Conyngham. Sydney Machine Learning Founder. Co-creator StarAi, Director DSAi.

    This motherfucker probably hasn’t ever even seen a dog.

  • pHr34kY@lemmy.worldEnglish
    2 months

    Did ChatGPT mention that a vaccine won’t work if the dog already has cancer?

  • notthebees@reddthat.comEnglish
    2 months

    That’s actually super neat. They really buried the lead with that article headline.

    • baatliwala@lemmy.worldEnglish
      2 months

      I swear they’re no different than the stereotypical conservative these days, absolutely repulsed by the fact that something good can come out of new ways of working