• NuXCOM_90Percent@lemmy.zip
    link
    fedilink
    English
    arrow-up
    4
    arrow-down
    5
    ·
    2 days ago

    I’m rapidly feeling like it’s pointless to bother learning when you can badly cobble something together with a chatbot and the managers of the world will ejaculate themselves dry about how good robots are, even when the bloody thing barely functions.

    Rhetorical question: How much of your decade of development has been in a professional capacity

    That has ALWAYS been true. A barely functioning Proof Of Concept has always been sexy. Someone has an idea, they make a barely functioning example of it working (often depending on stack overflow and asking others for help), show it to Management, and get money. With Management often thinking how they can either rapidly patent something in there or sell it off to a larger company.

    Nothing there is new aside from “AI” replacing “ask Stack Overflow”.

    And, just to be clear, that was also true in the hobbyist space. Think about how often you saw an article like “someone recreated PT in Unreal Engine!!!” (not to mention PT itself being the kind of project you give a new hire to learn the toolchain but…). Same with all those emulators that “added VR” and so forth. They are cool concepts that tend to not go anywhere or…

    Once a POC becomes a Product? That is where knowledge matters. You no longer want the answer someone shat out while waiting for a belle claire video to download. You need to actually define your corner cases, improve performance, and build out a roadmap.

    And… that ALSO isn’t about learning new tools and tech. A lot of that comes out of it, but that is where the difference between “computer programmer” and "software engineer’ comes into play. Because it becomes an engineering problem where you define and implement testing frameworks and build out the gitlab issues and so forth.

    Like, a LOT of dumbfucks try to speedrun their way to management because it is more money. But the reality is that a good Engineer SHOULD become a manager as they “grow up”. Because you need people with technical ability to have a say in building out that roadmap and in allocating resources to different issues. Optimally you still get to code a lot (I am a huge fan of middle management in that regard) but… yeah.

    • Die4Ever@retrolemmy.com
      link
      fedilink
      English
      arrow-up
      4
      ·
      2 days ago

      Unreal PT was actually really good lol. I had a friend play it since they didn’t have access to the original PT and they had a great time with it.

      But yeah most of those Unreal Engine “remakes” are just a couple of graphics assets with no real gameplay or content

    • Rayquetzalcoatl@lemmy.world
      link
      fedilink
      English
      arrow-up
      4
      ·
      2 days ago

      I see what you’re saying, but I’m not talking about proof of concepts. I’m talking about “fully fledged” Frankenstein apps that get cobbled together by cowboys. Documentation written by ChatGPT that is full of hallucinations. Managers love that stuff because the thing they’ve asked for works but nothing outside of that one thing works, which doesn’t matter because they’re not testing it.

      I’m not talking about small proof of concepts. I was referring to myself in a professional capacity as a developer; I’ve been a web developer full time since 2015.

      • NuXCOM_90Percent@lemmy.zip
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        2
        ·
        2 days ago

        Managers love that stuff because the thing they’ve asked for works but nothing outside of that one thing works, which doesn’t matter because they’re not testing it.

        Yeah. That is a POC. It is what you use to get funding, have lawyers write up a patent, or shop around the company

        • Rayquetzalcoatl@lemmy.world
          link
          fedilink
          English
          arrow-up
          3
          ·
          2 days ago

          That is not what I’m saying - the situation I’m describing is the situation I’m currently in: I work for a small web agency, we have the agency owner, the project manager, and the development lead as our “management”.

          A client asks for something, the agency owner says yes, and then the development lead cobbles something together over the course of a few hours with results from ChatGPT or Claude.

          The thing works, but only for that specific request and cannot handle edge cases, and he doesn’t know how it works nor how to extend it, so he cobbles on more ChatGPT or Claude results.

          The management team love it, but it’s just mountains of technical debt piling up.

          • NuXCOM_90Percent@lemmy.zip
            link
            fedilink
            English
            arrow-up
            1
            arrow-down
            2
            ·
            2 days ago

            A client asks for something, the agency owner says yes, and then the development lead cobbles something together over the course of a few hours with results from ChatGPT or Claude.

            Again… that is a POC.

            The thing works, but only for that specific request and cannot handle edge cases, and he doesn’t know how it works nor how to extend it, so he cobbles on more ChatGPT or Claude results.

            So… what you are saying is they make something specifically meeting the requirements given to them by the client with no intention of long term support? And that, in the event that you provide long term support, the skillset required drastically changes? Possibly to a more Software Engineering based one?

            • Rayquetzalcoatl@lemmy.world
              link
              fedilink
              English
              arrow-up
              3
              arrow-down
              1
              ·
              2 days ago

              It’s not a proof of concept, or an MVP - I’m saying it’s what is given to the client as a full, finished solution.

              What I’m saying is that the “this was built by AI!” effect is so strong that copy/pasting ChatGPT results together with no forethought or understanding is miserable and brings only technical debt - but that that’s irrelevant to management, because they’re impressed by the robot and don’t want to “fall behind” other agencies.