• kescusay@lemmy.world
    link
    fedilink
    English
    arrow-up
    14
    ·
    2 days ago

    The trick to using an AI agent effectively is already knowing exactly what you want, typing the request out in excruciating detail, and being a good developer who properly reviews code so you catch all the errors and repetition the AI agent will absolutely include.

    So… Yeah. 100% agree. AI agents are useful, but impossible to use if you aren’t already skilled with code.

    • residentmarchant@lemmy.world
      cake
      link
      fedilink
      English
      arrow-up
      4
      arrow-down
      1
      ·
      2 days ago

      Agree completely! The people saying “LLMs don’t produce good code” are using prompts like “build a feature that does X, Y and Z”

      Good prompting with current-quality LLMs needs to look like “create a function that take in params A and B and produces an output of C”

      It’s still faster than hand writing the code since the agent will refactor as it goes and break things down into manageable, small functions, but you have to tell it to do that.

    • kkj@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      1
      ·
      edit-2
      1 day ago

      Copilot does a good job of typing out things like imports that should really be loops but can’t be. Sure, I could easily write a Python or Bash script to do it, but that would take 5-10 minutes and just pressing Tab 20 times takes a lot less. I just have to read each line to make sure it didn’t hallucinate any files that I don’t actually have.