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
    link
    fedilink
    arrow-up
    68
    ·
    edit-2
    11 months 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
      ·
      11 months 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
        link
        fedilink
        arrow-up
        26
        ·
        edit-2
        11 months 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
        arrow-down
        1
        ·
        11 months 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.

              • t3rmit3@beehaw.org
                cake
                link
                fedilink
                arrow-up
                4
                ·
                11 months ago

                I actually did that for some code, and it did work.

                I asked chatgpt to write me a prompt that would make chatgpt write a recursive function for uploading files and all files in subdirectories to a server as “multipart forms”, because when I asked it to modify my code originally it was just giving me a do-while loop, whereas I wanted a recursive function.

                I kept changing my prompts to try to phrase “write a recursive function” differently, and instead the prompt that chatgpt gave me explicitly told it not to use non-recursive logic. Weirdly, forbidding it from using non-recursive logic actually made it finally give me the proper function.

        • jarfil@beehaw.org
          link
          fedilink
          arrow-up
          10
          arrow-down
          1
          ·
          11 months ago

          If we can have SEO be a thing, then we can have “Prompt Engineer” be a thing…

          Actually, I’ve been a “Google Search Prompt Engineer” for like 20 years already 🤷