• ipkpjersi@lemmy.ml
      link
      fedilink
      English
      arrow-up
      3
      arrow-down
      6
      ·
      21 hours ago

      Is it still slop when an experienced developer uses AI to assist them with development and thoroughly reviews it and is ultimately responsible for the code?

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

        Eventually, yes.

        LLM’s require expertise to supervise. Expertise requires practice to built and maintain. If you have the LLM do things, you’re getting less practice in, which slowly degrades your expertise, which in turn degrades your ability to supervise the LLM.

        • ipkpjersi@lemmy.ml
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 hour ago

          It degrades your experience writing code, but it doesn’t degrade your experience reviewing code. Those are two different skillsets.

      • sparky@lemmy.federate.cc@lemmy.federate.cc
        link
        fedilink
        English
        arrow-up
        3
        arrow-down
        1
        ·
        edit-2
        20 hours ago

        No, but then, a developer who is responsibly using LLMs as a tool in their IDE to do boilerplate generation kind of stuff, is fundamentally not vibe coding and thus it isn’t the case that the project is AI generated, just like AI assisted.

        I don’t really consider them the same use case. The former is low or no effort while the latter is a skilled person effectively doing the same types of offloading they used do with e.g. Jetbrains tools, just levelled up.

        Personally I think there is a huge difference between “generate the backing implementation of this interface/protocol I wrote” and “hey Claude make me a game”