Legal Scamming

>We present GPQA, a challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. We ensure that the questions are high-quality and extremely difficult: experts who have or are pursuing PhDs in the corresponding domains reach 65% accuracy (74% when discounting clear mistakes the experts identified in retrospect), while highly skilled non-expert validators only reach 34% accuracy, despite spending on average over 30 minutes with unrestricted access to the web
They're basically telling investors these bots, which can barely count or solve basic logic problems, are smarter than most high IQ university students and almost as smart as expert PhDs. How is this anything but a blatant scam and why is this allowed?

Black Rifle Cuck Company, Conservative Humor Shirt $21.68

UFOs Are A Psyop Shirt $21.68

Black Rifle Cuck Company, Conservative Humor Shirt $21.68

  1. 4 weeks ago
    Anonymous

    >investors
    rich idiots ripe for the plucking.
    you have money, you dont need more you greedy fricks. this isnt a scam its doing Gods Work.

  2. 4 weeks ago
    Anonymous

    >despite spending on average over 30 minutes with unrestricted access to the web
    >for 448 questions
    is that supposed to be a lot of time? that just makes it sound like the test takers got bored and didn't want to put in 50 hours of research to get a high score on a test that doesn't matter. in comparison, the AI has the equivalent of one gorillion hours of internet access

  3. 4 weeks ago
    Anonymous

    Or you didn't need a very high IQ to memorize stuff from college and get good grades? Are you moronic?

    • 4 weeks ago
      Anonymous

      You're American, aren't you? I can tell by your level of reading comprehension and your impulsive, apelike, incoherent spergouts.

      • 4 weeks ago
        Anonymous

        Dumbo the ability to answer multiple choice questions isn't correlated to IQ and that's all there is to it. Besides, believing that you need to be extra smart to get a PhD is peak midwit thinking

        • 4 weeks ago
          Anonymous

          Your reading comprehension fails on multiple levels, the most obvious of which is that I clearly acknowledged the fallacy of judging the bot's intelligence based on such a benchmark. Your post is moronic for several other reasons also, but this is enough to demonstrate that you are a barely sentient ape and that nothing you say warrants consideration.

  4. 4 weeks ago
    Anonymous

    Look at the actual dataset, the questions are designed to be answered by LLM. Using five-shot on questions like "which of the following is correct", is just a small part of the current AI scam.

  5. 4 weeks ago
    Anonymous

    >AGI in two more weeks

    • 4 weeks ago
      Anonymous

      use the API.
      the "free" model on the mainpage is hardlocked to use the smallest amount of computing power and tokens. It only reads 2 r because at the 3rd r the computing power is already used up and it just makes shit up.

      AGI is already possible, its just a question of computing power

      • 4 weeks ago
        Anonymous

        The level of psychotic cope is insane.
        >AGI is already possible
        How do you justify this claim when you don't have a single example of an "AI" capable of so much as basic reasoning?

        • 4 weeks ago
          Anonymous

          because no one is willing to invest 1 trillion dollars for a "conscious" chatbot.
          AGI is ready in theory. If you are willing to build 10 computer farms just for a chatbot go ahead.

          For the rest we need to wait until someone is willing to go all in

          • 4 weeks ago
            Anonymous

            >AGI is ready in theory.
            And yet you have not a shred of evidence to justify this claim. Curious.

          • 4 weeks ago
            Anonymous

            yea, i dont have a private computer farm the size of luxemburg

          • 4 weeks ago
            Anonymous

            You don't have anything at all besides unsubstantiated fantasies. You keep screeching "2 trillion more layers", "2 more years", "2 more city-sized corporate datacenters" etc. yet no existing architecture has shown potential for genuine intelligence; to the contrary, we can see that all current methods optimize for the wrong thing, that is cheating and fooling the human observer in increasingly more sophisticated ways rather than learning how to reason.

      • 4 weeks ago
        Anonymous

        >AGI is already possible
        >>>/x/

      • 4 weeks ago
        Anonymous

        >AGI is already possible, its just a question of computing power
        i cant believe R/Technology/ just leaked this!!!

    • 4 weeks ago
      Anonymous

      you think that's dumb? I gave it binary instructions and when all it had to start out with was COPY the instructions over, it already fricked up and literally wasn't able to COPY OVER the function binary. Just read the end of the first line between chatgpt's instructions and mine. It's literally moronic and a scam.

      • 4 weeks ago
        Anonymous

        That means it's working.

      • 4 weeks ago
        Anonymous

        This is 4o btw, not even 3.5

        • 4 weeks ago
          Anonymous

          Try regular 4. 4o is braindamaged.

          • 4 weeks ago
            Anonymous

            Same crap, I have tried it all. It keeps making hilarious dumb mistakes.

  6. 4 weeks ago
    Anonymous

    what parameters did they use?
    what exact model?
    temperature?
    tokens?
    randomization?
    number of tries?
    system messages?

    hello?

  7. 4 weeks ago
    Anonymous

    You can’t stop what’s coming.

    • 4 weeks ago
      Anonymous

      Call me back when you have an "AI" capable of solving 2nd grade level math/logic riddles.

  8. 4 weeks ago
    Anonymous

    >these bots, which can barely count or solve basic logic problems, are smarter than most high IQ university students and almost as smart as expert PhDs.
    is it a scam when it's true? i don't know if you've noticed, but universities aren't admitting or attracting the best of the best these days.

    • 4 weeks ago
      Anonymous

      >is it a scam when it's true?
      I know you're too jealous and butthurt to admit it, but being able to count and to reason at or above primary school level is, in fact, a necessary condition to become a PhD in any of the subjects listed in the GPQA.

  9. 4 weeks ago
    Anonymous

    >They're basically telling investors these bots, which can barely count or solve basic logic problems, are smarter than most high IQ university students and almost as smart as expert PhDs. How is this anything but a blatant scam and why is this allowed?
    how is your reading comprehension that bad?

    • 4 weeks ago
      Anonymous

      Sorry about your extreme mental illness.

      • 4 weeks ago
        Anonymous

        >t. le 36% face

        • 4 weeks ago
          Anonymous

          Uh oh. I broke the bot.

Your email address will not be published. Required fields are marked *