This is what drives me nuts. It’s pure laziness that drives this shit.

You know what would actually improve productivity? Allowing use of ad blockers so I can study work related content or instructional videos without a million ads. An email client that isn’t a piece of shit like outlook. A chat program that is not the horrendous teams. A cloud storage solution that is not one drive. Everything I’ve mentioned could/would be improved by open source implementation, but they think throwing an llm at it will improve productivity. Its just laziness. AND the fact that millions are spent on a business llm account per year is utterly stupid. I’ve not seen one scenario at our work where an llm has actually improved productivity. Sure, maybe some have used it for fixing bad grammar or writing an email they were too lazy to come up with, but to me if you are that dumb or lazy, you don’t deserve a good job.

To add. I’ve actually tried using the llm for certain things, and it maybe helped 5% of the time. Every other time it was wrong, or wasted more time than it took to actually do the work. Its scary how many blindly trust it and think they are working efficiently, while those of us in the background fix all the shit they screwed up because they were lazy.

Maybe something good will happen and the dumb will be easier to weed out because of this. That is one potential upside.

  • vrek@programming.dev
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    4
    ·
    6 days ago

    The one case where it helps is in formatting. Like my boss often asks like “here’s a excel file of 200 serial numbers, can you get the info from the database for them?” and yes but I need to erase the carriage return and an extra space from each so it will be 2 hours… Or I ask copilot “take this query and add in a where clause for each of these serial numbers” and it works.

    Now asking it write that query is probably failure but simple formatting tends to work.

    • NightFantom@slrpnk.net
      link
      fedilink
      arrow-up
      12
      ·
      6 days ago

      Last time a colleague asked chatgpt to format a document, a few ids got removed and new ones got hallucinated in. Still laugh whenever I remember, thanks for that 🤣

      • Log in | Sign up@lemmy.world
        link
        fedilink
        arrow-up
        1
        ·
        4 days ago

        I have a similar story, but it got worse and worse with the lies as it got through the table. I fought it for an hour, then I wrote a script instead.

    • Paid in cheese@lemmings.world
      link
      fedilink
      English
      arrow-up
      13
      ·
      6 days ago

      There are so many better tools out there for sanitizing data like that. Tools that don’t run the risk of inserting hallucinations into your work. Granted, they do take effort to learn (Regular Expressions would be the obvious one). If it needs to be right or even has to be right, I can’t imagine risking it to an LLM / GAN tool.

    • shalafi@lemmy.world
      link
      fedilink
      English
      arrow-up
      3
      ·
      5 days ago

      I would, and have used, PowerShell for that. No doubt AI is faster but I’d trust my script output.