Researchers found that ChatGPT’s performance varied significantly over time, showing “wild fluctuations” in its ability to solve math problems, answer questions, generate code, and do visual reasoning between March and June 2022. In particular, ChatGPT’s accuracy in solving math problems dropped drastically from over 97% in March to just 2.4% in June for one test. ChatGPT also stopped explaining its reasoning for answers and responses over time, making it less transparent. While ChatGPT became “safer” by avoiding engaging with sensitive questions, researchers note that providing less rationale limits understanding of how the AI works. The study highlights the need to continuously monitor large language models to catch performance drifts over time.

  • calculuschild@lemm.ee
    link
    fedilink
    arrow-up
    56
    ·
    edit-2
    1 year ago

    My understanding is this claim is basically entirely false. The tests done by these researchers had some glaring errors that when corrected, show gpt-4 is getting slightly better at math, if anything. See this video that describes some of the issues: https://youtu.be/YSokS2ivf7U

    TL;DR The researchers gave new GPT questions from two different pools. It’s no surprise they got worse answers.

    • Blóðbók@slrpnk.net
      link
      fedilink
      arrow-up
      19
      ·
      1 year ago

      You shouldn’t need to be a prompt engineer just to get answers to math questions that are not blatantly wrong. I believe the prompts are included in the paper so that you don’t have to guess if they were badly formatted.

      • calculuschild@lemm.ee
        link
        fedilink
        arrow-up
        26
        ·
        edit-2
        1 year ago

        The problem is they aren’t comparing apples to apples. They asked each version of GPT a different pool of questions. (Edited my post to make this clear).

        Once you ask them the same questions, it becomes clear that ChatGPT isn’t getting worse at math, because it has been terrible all along.

      • Mars@beehaw.org
        link
        fedilink
        arrow-up
        22
        ·
        1 year ago

        “Prompt Engenieer” is one of the funniest thinks that have happened in the recent history of the world.

        “Learn to ask questions to a prediction algorithm and get rich! Is the work of the future! Software engineers and writers will lose their jobs, but asking questions is an evergreen field!”

        Dude, if the algorithm only understand correctly formatted input is a parser. We have those.

  • stravanasu@lemmy.ca
    link
    fedilink
    English
    arrow-up
    31
    ·
    edit-2
    1 year ago

    For me it’s like using a coffee machine as a stopwatch, and then complaining that it doesn’t always give the exact time lapsed.

    • nottheengineer@feddit.de
      link
      fedilink
      English
      arrow-up
      4
      ·
      1 year ago

      If it’s a coffee machine that’s so advanced it was uninaginable a decade ago, you’d expect it not to perform worse over time.

      • stravanasu@lemmy.ca
        link
        fedilink
        English
        arrow-up
        6
        ·
        1 year ago

        My point was that a coffee machine is designed to make coffee, not to keep track of time. Maybe it always takes roughly the same amount of time to make a coffee, and so someone uses it as a proxy stopwatch. But it can very well suddenly take more or less time, without anything being wrong about it – maybe different coffee brands, cleaned pipes, or whatnot.

        ChatGPT is an algorithm designed to parrot language, not to perform mathematical reasoning based on logic rules.

        • jarfil@beehaw.org
          link
          fedilink
          English
          arrow-up
          4
          ·
          edit-2
          1 year ago

          ChatGPT is an algorithm designed to parrot language, not to perform mathematical reasoning based on logic rules.

          Mathematical language is a language, ChatGPT has been shown to come up with relationship between very distant elements of language that were not present in the training data… so there is nothing stopping it from, given enough mathematical training data, coming up with whatever relationships we call “logical rules”.

          • stravanasu@lemmy.ca
            link
            fedilink
            English
            arrow-up
            5
            ·
            edit-2
            1 year ago

            Mathematical language is a language, but mathematics is not just a language. It is a structure with internal rules that are not determined by pure convention (as natural languages are). We could internationally agree from tomorrow to call “blue” whatever it’s now called “red” and vice versa, but we couldn’t agree to say that “2 + 2 = 5”, because that would lead to internal inconsistencies (we could agree to use the symbol “5” for 4, but that’s a different matter).

            This is also related to a staple of science: that scientific and mathematical truth is not determined by a majority vote, but by internal consistency. Indeed modern science started with this very paradigm shift. Quoting Galilei:

            But in the natural sciences, whose conclusions are true and necessary and have nothing to do with human will, one must take care not to place oneself in the defense of error; for here a thousand Demostheneses and a thousand Aristotles would be left in the lurch by every mediocre wit who happened to hit upon the truth for himself.

            If we want to train an algorithm to infer rules from language, we need to give samples of language where the rules are obeyed strictly (and yet this may not be enough). Otherwise the algorithm will wrongly generalize that the rules aren’t strict (in fact it’ll just see a bunch of mutually inconsistent examples). Which is what happens with ChatGPT.

            Edit: On top of this, Gödel’s theorem and other related theorems have shown that mathematical reasoning cannot be reduced to pure symbol manipulation, Hilbert’s unfulfilled dream. So one can’t infer mathematical reasoning from language patterns. Children learn reasoning not only through language training, but also through behaviour training (this was pointed out by Turing). This is why large language models have intrinsic limitations in what they can achieve and be used for.

        • elevenant@discuss.tchncs.de
          link
          fedilink
          English
          arrow-up
          3
          ·
          1 year ago

          You are not wrong, but I think public perception is different. It doesn’t help, that OpenAI is pushing their models as problem solvers:

          GPT-4 can solve difficult problems with greater accuracy, thanks to its broader general knowledge and problem solving abilities. (https://openai.com/gpt-4)

  • Big P@feddit.uk
    link
    fedilink
    arrow-up
    13
    ·
    1 year ago

    I think this might be what stops AI from taking over as much as people fear. If I was a business owner I wouldn’t want to put my trust in a black box if I can pay someone to ensure it works exactly to my specification

    • PeepinGoodArgs@reddthat.com
      link
      fedilink
      arrow-up
      13
      ·
      1 year ago

      As someone getting an MBA that hates the idea of labor being displaced by AI, if I were an unethical business owner that treated labor as a cost to minimize, I’d use AI to generate content that’s “good enough” and use fewer people to make it exactly to my specification.

      • Fenzik@lemmy.ml
        link
        fedilink
        arrow-up
        8
        ·
        1 year ago

        You uh… you might have chosen the wrong field if you hate displacing labour

      • django@discuss.tchncs.de
        link
        fedilink
        arrow-up
        8
        ·
        1 year ago

        You know, I wouldn’t care about being replaced by a machine, as long as I get UBI. Then I could just do what I like to do and wouldn’t need to care whether I actually make money with it.

        • Chahk@beehaw.org
          link
          fedilink
          arrow-up
          3
          ·
          1 year ago

          That’s not how UBI is supposed to work. You would certainly have enough time to do what you like, just not the resources. Any money you’d get would only cover the absolute necessities like shelter and food.

          • snowbell@beehaw.org
            link
            fedilink
            arrow-up
            1
            ·
            edit-2
            1 year ago

            According to who? Who defines what a “basic necessity” is? It could easily be argued that hobbies are a necessity.

      • lemmy689@lemmy.sdf.org
        link
        fedilink
        arrow-up
        3
        ·
        1 year ago

        I think that’s what part of the Hollywood writers strike is about. AI generating “good enough” scripts, and studios shelling a few peanuts for some writers to finalize them.

  • PoisonedPrisonPanda@discuss.tchncs.de
    link
    fedilink
    arrow-up
    7
    ·
    1 year ago

    I apologize for my naivity.

    but could openAI just introduce a flag into the decoder to highlight math questions and ports/transforms those math questions into a simple bash script to calculate the result instead of letting the LLM nodes “calculate” the formula?

    I mean this would like straightforward give correct results. ChatGPT has a similar issue with counting as its nodes do not get the numerics. however a pc is capable of that. it would just rely on the encoder for parsing the question, and not going further the GPT route.

  • Karlos_Cantana@sopuli.xyz
    link
    fedilink
    arrow-up
    7
    ·
    1 year ago

    I’ve found it making up “facts” when I query it. I thought I was doing something wrong, but apparently, it’s just changing the way it works for some reason.

    • FIash Mob #5678@beehaw.org
      link
      fedilink
      arrow-up
      12
      ·
      edit-2
      1 year ago

      Same. Now I’m only using search engines that don’t have it.

      It’s not changing the way it works. It’s making up shit when it doesn’t know.

      • Scrubbles@poptalk.scrubbles.tech
        link
        fedilink
        arrow-up
        12
        ·
        1 year ago

        And that’s how AI works, it’s all probability. It’s not answering 2+2, there’s a probability that the answer is 4 and it chooses that. If something convinces it that it should be 5 it’ll start answering 5

        • Rhaedas@kbin.social
          link
          fedilink
          arrow-up
          13
          ·
          1 year ago

          That’s how language models work. It’s grouped into AI as is so many things, but it’s not AGI. It could open the doors to AGI as a component, but isn’t actually thinking about its answers. And those probabilities are driven by training reinforcement which includes the bias of giving an answer the human will receive well. Of course it’s going to “lie” or make up things if that improves the acceptance of the answer given.

          • aperson@beehaw.org
            link
            fedilink
            arrow-up
            7
            ·
            1 year ago

            The best description I’ve heard to give to most people is that llms knows what the right answer looks like, not what it is.

      • thejml@lemm.ee
        link
        fedilink
        arrow-up
        5
        ·
        1 year ago

        If I wanted that I could just ask my daughter. She makes up shit all the time when she doesn’t actually know.

        • hypevhs@beehaw.org
          link
          fedilink
          English
          arrow-up
          10
          ·
          1 year ago

          According to the Japanese zodiac, people born in May 1994 would have the zodiac sign of the Snake.

          Expect it’s Dog, not Snake. Bing thinks it’s Ox. How did the entire field of AI go from surprisingly accurate to utterly useless in the span of under a year? I have no idea what benefits you personally see in this site.

          • PeepinGoodArgs@reddthat.com
            link
            fedilink
            arrow-up
            1
            ·
            1 year ago

            Oh boy. I do research on it for various things. Florida released some laws changing alimony and I researched it via Perplexity to understand what the problem was. It worked. I understood the issue.

            Or carbon capture technology.

            In any case, I do look directly at the sources. Perplexity.ai is useful for framing a topic, getting the gist of it, but for being sure I know wtf is going on, I personally need to look at the sources.

            • Very_Bad_Janet@kbin.social
              link
              fedilink
              arrow-up
              3
              ·
              edit-2
              1 year ago

              Thanks for this reply. That’s probably the best way to use LLMs - general definitions or framing / summarizing of issues. And then always check the sources to make sure it was accurate. I’ve played around with ChatGPT and Bard and I think my mistake has been to be a little too granular or specific in my prompts. In most cases it produced results that were inaccurate (ETA: or flat out demonstrably wrong) or only fulfilled a part of the prompt.

              • PeepinGoodArgs@reddthat.com
                link
                fedilink
                arrow-up
                1
                ·
                1 year ago

                the best way to use LLMs - general definitions or framing / summarizing of issues. And then always check the sources to make sure it was accurate.

                I agree. The criticism that they’re not accurate kinda misses the point of LLMs being tools. It’d be like complaining that a hammer doesn’t jam the nail in all the way after the first stroke. Hit it again…and maybe try hitting it straight this time instead of at an angle. It’s an iterative process that can be self-correcting when done thoughtfully.

        • backpackn@lemmy.ml
          link
          fedilink
          arrow-up
          1
          ·
          edit-2
          1 year ago

          Was gonna say this too, it’s a great one for fact-checking. Sometimes it won’t include a source and make something up, just watch out for those.

  • Gnubyte@lemdit.com
    link
    fedilink
    arrow-up
    5
    ·
    1 year ago

    To be honest I noticed a drop in quality of code generation via prompt by ChatGPT.

    Still useful. Especially for boilerplate nonsense getting projects started. But it’s ability to understand complexities in code dropped drastically.