• Ephera@lemmy.ml
    link
    fedilink
    English
    arrow-up
    5
    ·
    5 hours ago

    You not using any libraries? We have like 1100 animals in our software at work (that we know of).

  • rozodru@piefed.social
    link
    fedilink
    English
    arrow-up
    27
    ·
    11 hours ago

    Claude: “It’s a known issue”

    ChatGPT: “Ah - now I see the what the problem is you’re trying to do such and such” after telling me to do such and such

    Gemini: “I see you’re trying to bake a lettuce cake with chocolate chip cookies”

    • taxon@lemmy.world
      link
      fedilink
      English
      arrow-up
      6
      ·
      7 hours ago

      More like, Anymodel: “You’re absolutely right! What the hell is this?”

  • myfunnyaccountname@lemmy.zip
    link
    fedilink
    arrow-up
    10
    ·
    11 hours ago

    Welcome to corporate America. If it works, ship it. Fuck it. It’s not your personal pet project. Give back as much as they care about you.

    • b0ber@lemmy.worldOP
      link
      fedilink
      arrow-up
      6
      arrow-down
      4
      ·
      15 hours ago

      Still useful to this day that’s true. But I also like to gamble on my code and argue with AI

        • b0ber@lemmy.worldOP
          link
          fedilink
          arrow-up
          5
          arrow-down
          1
          ·
          14 hours ago

          Well, to be honest it’s kind of like that 90% of the time. You don’t actually become more productive but end up cleaning trash most of the time. Anyway, the hardest part is connecting all the pieces properly and understanding how everything fits, handling edge cases, and so on. Writing the code itself is actually the easy part.

          • SpaceNoodle@lemmy.world
            link
            fedilink
            arrow-up
            18
            arrow-down
            3
            ·
            14 hours ago

            That’s all part of writing it. Typing gibberish you don’t understand or that doesn’t work doesn’t count.

            • b0ber@lemmy.worldOP
              link
              fedilink
              arrow-up
              6
              ·
              14 hours ago

              Not necessarily, even before having all these models, I spent most of my time thinking rather than coding. The coding part is just the final nail, that’s my approach at least.

              • Ignotum@lemmy.world
                link
                fedilink
                arrow-up
                2
                arrow-down
                1
                ·
                10 hours ago

                Same here really, i plan the structure, then have ai throw it together, then i clean up and build off of that

                I can’t say i particularly miss writing the mindless boilerplate stuff

                • IcyToes@sh.itjust.works
                  link
                  fedilink
                  arrow-up
                  1
                  ·
                  3 hours ago

                  Surely missing that process, you forget error handling. As you write, its consistently, what if this is null, what if the input is like x. There are many questions you should ask yourself and if you AI it, you’ll forget.

  • ominous ocelot@leminal.space
    link
    fedilink
    arrow-up
    4
    ·
    edit-2
    8 hours ago

    It is called sampling and recombination, which is perfectly fine. The only metric should be that the final product masters the testing parkour. And that parkour better be worth its name. :)