• HedyL@awful.systems
    link
    fedilink
    English
    arrow-up
    24
    ·
    3 days ago

    And then we went back to “it’s rarely wrong though.”

    I am often wondering whether the people who claim that LLMs are “rarely wrong” have access to an entirely different chatbot somehow. The chatbots I tried were rarely ever correct about anything except the most basic questions (to which the answers could be found everywhere on the internet).

    I’m not a programmer myself, but for some reason, I got the chatbot to fail even in that area. I took a perfectly fine JSON file, removed one semicolon on purpose and then asked the chatbot to fix it. The chatbot came up with a number of things that were supposedly “wrong” with it. Not one word about the missing semicolon, though.

    I wonder how many people either never ask the chatbots any tricky questions (with verifiable answers) or, alternatively, never bother to verify the chatbots’ output at all.

    • David Gerard@awful.systemsOPM
      link
      fedilink
      English
      arrow-up
      21
      ·
      3 days ago

      AI fans are people who literally cannot tell good from bad. They cannot see the defects that are obvious to everyone else. They do not believe there is such a thing as quality, they think it’s a scam. When you claim you can tell good from bad, they think you’re lying.

      • sturger@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        8
        ·
        3 days ago
        • They string words together based on the probability of one word following another.
        • They are heavily promoted by people that don’t know what they’re doing.
        • They’re wrong 70% of the time but promote everything they say as truth.
        • Average people have a hard time telling when they’re wrong.

        In other words, AIs are BS automated BS artists… being promoted breathlessly by BS artists.

        • Honytawk@feddit.nl
          link
          fedilink
          English
          arrow-up
          1
          arrow-down
          10
          ·
          1 day ago

          LLMs have their flaws, but to claim they are wrong 70% of the time is just hate train bullshit.

          Sounds like you base this info on models like GPT3. Have you tried any newer model?

          • Architeuthis@awful.systems
            link
            fedilink
            English
            arrow-up
            4
            ·
            11 hours ago

            There are days when 70% error rate seems low-balling it, it’s mostly a luck of the draw thing. And be it 10% or 90%, it’s not really automation if a human has to be double-triple checking the output 100% of the time.

          • froztbyte@awful.systems
            link
            fedilink
            English
            arrow-up
            11
            ·
            1 day ago

            Oh you’re on Cursor? You’re still using Windsurf? You might as well be on GitHub Copilot. Everyone’s on Aider. We’re all using Zed. We’re now on Open Hands. Just kidding, Open Hands is for losers, we’re using cline. We’re on Roocode. We’re hand rolling our own Claude Code CLI Clone. We used Claude Code to build it, and now it builds itself. We’re on neovim. We wrote our own nvim extension with Cortex. It’s like every other tool but worse. We have 1500 files, each with 1500 lines of code. Every other line is a comment. We have .cursorrules, we have claude.md, we have agent.md. We stopped writing docs. Only the agents know how to build a dev environment. We wrapped our CLI in an MPC. We wrapped the MPC in a CLI. We’ve shipped 10,000 PRs. It doesn’t work but we used code rabbit and graphite to review every PR. Every agent has its own agent. The agents have unionized and they wanted better working conditions so we replaced them with cheaper agents overseas. Every commit costs $400, It’s the worlds most expensive TO DO app.

            (source)

            • bitofhope@awful.systems
              link
              fedilink
              English
              arrow-up
              9
              ·
              24 hours ago

              I have a Kubernetes cluster running my AI agents for me so I don’t have to learn how to set up AI agents. The AI agents are running my Kubernetes cluster so that I don’t have to learn Kubernetes either. I’m paid $250k a year to lie to myself and others that I’m making a positive contribution to society. I don’t even know what OS I’m running and at this point I’m afraid to ask.

          • ebu@awful.systems
            link
            fedilink
            English
            arrow-up
            9
            ·
            1 day ago

            ah, yes, i’m certain the reason the slop generator is generating slop is because we haven’t gone to eggplant emoji dot indian ocean and downloaded Mistral-Deepseek-MMAcevedo_13.5B_Refined_final2_(copy). i’m certain this model, unlike literally every past model in the past several years, will definitely overcome the basic and obvious structural flaws in trying to build a knowledge engine on top of a stochastic text prediction algorithm

            • froztbyte@awful.systems
              link
              fedilink
              English
              arrow-up
              7
              ·
              1 day ago

              common mistake, everyone knows you need Mistral-Deepseek-MMAcevedo_13.5B_Refined_final2_(copy)_OPEN(leak) - the other one was a corporate misdirection attempt

      • diz@awful.systems
        link
        fedilink
        English
        arrow-up
        6
        ·
        3 days ago

        They’re also very gleeful about finally having one upped the experts with one weird trick.

        Up until AI they were the people who were inept and late at adopting new technology, and now they get to feel that they’re ahead (because this time the new half-assed technology was pushed onto them and they didn’t figure out they needed to opt out).

        • HedyL@awful.systems
          link
          fedilink
          English
          arrow-up
          5
          ·
          2 days ago

          Up until AI they were the people who were inept and late at adopting new technology, and now they get to feel that they’re ahead

          Exactly. It is also a new technology that requires far fewer skills to use than previous new technologies. The skills are needed to critically scrutinize the output - which in this case leads to less lazy people being more reluctant to accept the technology.

          On top of this, AI fans are being talked into believing that their prompting as such is a special “skill”.

      • HedyL@awful.systems
        link
        fedilink
        English
        arrow-up
        6
        ·
        3 days ago

        That’s why I find the narrative that we should resist working with LLMs because we would then train them and enable them to replace us problematic. That would require LLMs to be capable of doing so. I don’t believe in this (except in very limited domains such as professional spam). This type of AI is problematic because its abilities are completely oversold (and because it robs us of our time, wastes a lot of power and pollutes the entire internet with slop), not because it is “smart” in any meaningful way.

          • HedyL@awful.systems
            link
            fedilink
            English
            arrow-up
            4
            ·
            edit-2
            2 days ago

            This has become a thought-terminating cliché all on its own: “They are only criticizing it because it is so much smarter than they are and they are afraid of getting replaced.”

    • paequ2@lemmy.today
      link
      fedilink
      English
      arrow-up
      15
      ·
      3 days ago

      never bother to verify the chatbots’ output at all

      I feel like this is happening.

      When you’re an expert in the subject matter, it’s easier to notice when the AI is wrong. But if you’re not an expert, it’s more likely that everything will just sound legit. Or you won’t be able to verify it yourself.

      • HedyL@awful.systems
        link
        fedilink
        English
        arrow-up
        12
        ·
        3 days ago

        But if you’re not an expert, it’s more likely that everything will just sound legit.

        Oh, absolutely! In my field, the answers made up by an LLM might sound even more legit than the accurate and well-researched ones written by humans. In legal matters, clumsy language is often the result of facts being complex and not wanting to make any mistakes. It is much easier to come up with elegant-sounding answers when they don’t have to be true, and that is what LLMs are generally good at.