I have tried, unsuccessfully, to get various AI models to create a script that will curl or wget the latest Ubuntu LTS desktop torrent, even should that LTS version update in the future (beyond 24.01.1 LTS). The purpose is that I would like to seed whatever the latest LTS torrent is and I don’t want to have to keep checking the Ubuntu page for updates, I want it automatic. I know that LTS is slow to change versions but I am annoyed that AI can’t just write a decent script for this.

I also have downloaded rtorrent as a command line and will deal with how to make sure the latest LTS is used, as opposed to the prior one, with a different script later, but that’s not what I’m trying to now.

I am not asking for a human to create this script for me. I am asking why AI models keep getting this so wrong. I’ve tried ChatGPT 4o, I’ve tried DeepSeek, I’ve tried other localized models, Reasoning Models. They all fail. And when I execute their code, and I get errors and show it to the models, they still fail, many times in a row. I want to ask Lemmy if getting an answer is theoretically possible with the right prompt or if AI just sucks at coding.

This shouldn’t be too hard to do. At https://www.releases.ubuntu.com, they list the releases. When curling the webpage, there’s a list of the releases with version numbers some with LTS. New versions are always larger numbers. At https://ubuntu.com/download/alternative-downloads, they list the torrents. Also, all release torrents for desktop are in the format https://www.releases.ubuntu.com/XX.XX/*desktop*.torrent. I’ve tried to teach these models this shit and to just create a script for me, holy shit it’s been annoying. The models are incredibly stupid with scripting.

I’m not a computer programmer or developer and am picking up more coding here and there just because I want to do certain things in linux. But I just don’t understand why this is so difficult.

So my question is, is there ANY prompt for ANY model that will output successful code for this seemingly easy task, or is AI still too stupid to do this?

  • over_clox@lemmy.world
    link
    fedilink
    arrow-up
    15
    ·
    2 months ago

    It took me more time to read this thread than it did for me to learn the basics of Bash scripting.

    Do like we did back in the day, RTFM

    • secretlyaddictedtolinux@lemmy.worldOP
      link
      fedilink
      arrow-up
      3
      arrow-down
      11
      ·
      edit-2
      2 months ago

      man bash is over 64,000 lines. we may also have different IQs, mine more suitable to asking profound philosophical questions to DeepSeek R1 (see https://lemmy.world/post/24838347 for example) and yours more suitable to learning bash scripting through electro-osmosis

      for everyone downvoting me, i have tried to get better at linux! don’t think of me as a lazy linux user, think of me as among the best windows users who switched to linux

  • AstridWipenaugh@lemmy.world
    link
    fedilink
    arrow-up
    4
    ·
    2 months ago

    AI certainly can do it. But here’s the thing with generative AI: the answer is only as good as the question you ask. If you don’t know exactly what to ask for and which details are important, the AI doesn’t know what you meant to ask and can’t infer that. AI usually does not pick up implied context that an experienced person would. A person would be able to make an educated guess about what you actually meant and answer that question.

    As someone with 20 years of programming experience, I would recommend against using AI to learn to program. You’re asking something that doesn’t actually know how to program to show you how to do it. From my experience with coworkers using AI, it doesn’t improve their work; it simply accelerates the rate at which they can produce low quality work.

    Once you’re more skilled than the AI, you can use it to speed up menial tasks, like generating boilerplate and stubbing things out. It absolutely will be wrong in some ways, and you need to be able to tell when it’s wrong and know how to fix it.

    • secretlyaddictedtolinux@lemmy.worldOP
      link
      fedilink
      arrow-up
      1
      arrow-down
      4
      ·
      2 months ago

      You’re probably right, but I’m not actually trying to learn to program. I learn programming and linux on an as needed basis, but I just keep learning more of it when I want to do stuff.

      If I can’t get AI to figure this out for me, I am not sure if it’s worth it to me to learn substantially more about bash programming just to bypass going to a webpage. It would probably be like 4 days of learning. From the page listing the torrents in a spreadsheet, I am guessing it would be easy to figure out how to list them, delete the ones without LTS in the name, order them by number and delete everything but the highest number. An LLM could probably help me with that task. I still don’t know if I have the IQ/insight amount to get it to a working level, so it’s also a gamble.

      It just seems like an LLM should be able to do all of this. This is sort of the quintessential trivial programming task that an LLM should be able to easily do.

  • edb_fyr@lemmy.world
    link
    fedilink
    arrow-up
    2
    ·
    2 months ago

    In my experience, Chat GPT is much better at the general programming patterns and concepts than specific API implementations. Especially less commonly used ones, where it often gives a solution that only looks reasonable but uses the totally wrong methods. I guess it’s similar with scraping the HTML of the Ubuntu releases page to get the latest torrent. It doesn’t know the exact HTML layout, so it guesses what is a likely one, even though it is wrong.

    • secretlyaddictedtolinux@lemmy.worldOP
      link
      fedilink
      arrow-up
      1
      arrow-down
      2
      ·
      2 months ago

      AI models are great for troubleshooting and understanding concepts or figuring out approaches but in actual implementation they seem weak to me. I’ve heard people saying how AI is replacing much of coding and I’m not really a coder but I can’t understand how that would be currently possible with the responses I see from AI,

  • TropicalDingdong@lemmy.world
    link
    fedilink
    arrow-up
    1
    ·
    2 months ago

    Early in I was really interested in what I could get chat got to do with minimal guidance. It really didn’t (at the time) have the ability to hold any kind of overarching “structure” or goal in mind. When it got something wrong, it would just wrongly keep trying the same thing.

    I’m still interested in what a local AI buddy could do on its own, but in terms actually ever accomplishing anything, I think it still very much needs its hands held.