• sheogorath@lemmy.world
      link
      fedilink
      arrow-up
      10
      ·
      7 months ago

      If your work is bleeding edge enough, even ChatGPT won’t be of help since it’s not in their training dataset.

      • locuester@lemmy.zip
        link
        fedilink
        English
        arrow-up
        8
        ·
        7 months ago

        Yeah and it won’t tell you that it hasn’t seen this pattern before. It will just make things up out of the blue which seem like they might be correct.

        Stay away from ChatGPT for bleeding edge things.

        • livingcoder@programming.dev
          link
          fedilink
          arrow-up
          2
          ·
          7 months ago

          It’s still useful when it’s wrong because it can give you the jist of what should be done. If it uses a library or function that doesn’t exist, you’ll still be informed as to what it was intending for the process at that point. I’ve often gone and just replaced the made-up code with custom code that does the same thing.

    • dantheclamman@lemmy.world
      link
      fedilink
      arrow-up
      3
      ·
      7 months ago

      It is nice to generate generalizable code examples, to give me clues how stuff works. I find that my work (marine biogeochemistry) is obscure enough that there’s a certain level where I am still on my own. Which is a good sign for my future employability!