• Contend6248@feddit.de
      link
      fedilink
      arrow-up
      26
      arrow-down
      1
      ·
      edit-2
      1 year ago

      True, when i respond with the exact problem it usually gets fixed, interestingly even explained why it failed.

      Great for learning

      • IzzyScissor@lemmy.world
        link
        fedilink
        arrow-up
        24
        ·
        1 year ago

        The only problem is that it’ll ALSO agree if you suggest the wrong problem.

        “Hey, shouldn’t you have to fleem the snort so it can be repurposed for later use?”

        You are correct. Fleeming the snort is necessary for repurposing for later use. Here is the updated code:

        • Rhaedas@kbin.social
          link
          fedilink
          arrow-up
          4
          ·
          1 year ago

          Models are geared towards seeking the best human response for answers, not necessarily the answers themselves. Its first answer is based on probability of autocompleting from a huge sample of data, and in versions that have a memory adjusts later responses to how well the human is accepting the answers. There is no actual processing of the answers, although that may be in the latest variations being worked on where there are components that cycle through hundreds of attempts of generations of a problem to try to verify and pick the best answers. Basically rather than spit out the first autocomplete answers, it has subprocessing to actually weed out the junk and narrow into a hopefully good result. Still not AGI, but it’s more useful than the first LLMs.

        • FaceDeer@kbin.social
          link
          fedilink
          arrow-up
          3
          ·
          1 year ago

          That’s not been my experience. It’ll tend to be agreeable when I suggest architecture changes, or if I insist on some particular suboptimal design element, but if I tell it “this bit here isn’t working” when it clearly isn’t the real problem I’ve had it disagree with me and tell me what it thinks the bug is really caused by.

  • Fades@lemmy.world
    link
    fedilink
    arrow-up
    30
    arrow-down
    7
    ·
    1 year ago

    Every time I hear this from one of my devs under me I get a little more angry. Such a meaningless statement, what are you gonna do, hand your pc to the fucking customer?

    • FaceDeer@kbin.social
      link
      fedilink
      arrow-up
      34
      ·
      1 year ago

      It’s not actually meaningless. It means “I did test this and it did work under certain conditions.” So maybe if you can determine what conditions are different on the customer’s machine that’ll give you a clue as to what happened.

      The most obscure bug that I ever created ended up being something that would work just fine on any machine that had at any point had Visual Studio 2013 installed on it, even if it had since had it uninstalled (it left behind the library that my code change had introduced a hidden dependency on). It would only fail on a machine that had never had Visual Studio 2013 installed. This was quite a few years back so the computers we had throughout the company mostly had had 2013 installed at some point, only brand new ones that hadn’t been used for much would crash when it happened to touch my code. That was a fun one to figure out and the list of “works on this machine” vs. “doesn’t work on that machine” was useful.

    • Platypus@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      25
      ·
      edit-2
      1 year ago

      doesn’t understand that this is a useful first step in debugging

      reacts with anger when devs don’t magically have an instant fix to a vague bug

      Yep, that’s a manager

  • Palacegalleryratio [he/him]@hexbear.net
    link
    fedilink
    English
    arrow-up
    6
    ·
    1 year ago

    You know what, this is on us. Chat GPT is just a prediction engine that tries to say what words it thinks follow a preceding prompt, and it’s looked at millions of examples (written by us) and it’s seen hapless clients and users complain about bugs and be told: “user error, works fine” so often chatGPT just thinks it’s just the culturally accepted polite response to a bug report, in the same way as responding to “thank you” with “you’re welcome”. This is a dark mirror on our profession.