• LostXOR@fedia.io
    link
    fedilink
    arrow-up
    5
    ·
    13 hours ago

    Replit? The online code compiler/interpreter website? I guess I’m not entirely surprised that they turned to slop.

  • cronenthal@discuss.tchncs.de
    link
    fedilink
    English
    arrow-up
    29
    ·
    19 hours ago

    Is this real? Because it sounds like some low effort satire about blindly trusting LLMs and the totally expected outcomes. Surely no one can be this naive.

    • cyrano@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      18
      ·
      18 hours ago

      User used the same db for prod and dev, user has no backup, LLM with db access is deleting it, user interacts with the LLM like it’s a human and ask it to apologize and follow promises of not doing it…… Oh and user doesn’t use git or any code linting/control.

      But yeah it is the llm fault /s What is scary is this is the tip of the iceberg. I foresee a lot of security problems in the future if software development goes that way.

      • PhilipTheBucket@quokk.auOP
        link
        fedilink
        English
        arrow-up
        5
        ·
        8 hours ago

        He actually did have a backup, because the company is only normal-stupid and not deliberate-stupid, they had a DB checkpoint he could roll back to.

        The LLM, of course, went with the path of least resistance once it started down the “oh no I fucked up” completion prompt, and claimed they had no such checkpoint.

        Don’t use LLM for fact things, kids.

    • .Donuts@lemmy.world
      link
      fedilink
      English
      arrow-up
      14
      ·
      18 hours ago

      Right? It’s crazy to think these kinds of people exist, but they are real and they make decisions for other people.

      “The [AI] safety stuff is more visceral to me after a weekend of vibe hacking,” Lemkin said. I explicitly told it eleven times in ALL CAPS not to do this. I am a little worried about safety now.

      • altkey@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        5
        ·
        18 hours ago

        I doubt their claim. How does LLM communicate directly to different systems in their infrastructure? What even promts it to act to begin with?

        Unless they went out of their way creating such interface for some reason, it is plain bullshit and human error, or a coverup by a skinbag CEO. He made screenshots of LLM taking the blame on itself that, as a concept, completely impossible, and we belive his lying ass lips. If only he asked it, at what stage AI is now, he could’ve lied better.

        • EpeeGnome@feddit.online
          link
          fedilink
          English
          arrow-up
          4
          ·
          10 hours ago

          It wasn’t the user’s infrastructure, it was the LLM company’s. The selling point is that it’s all integrated together for you. You explain what you want, the LLM not only codes it, but launches it too. Yes, his screen shots of the LLM “taking responsibility” are idiotic, but so many people don’t understand that LLMs don’t actually understand anything.