• RestrictedAccount
    link
    fedilink
    arrow-up
    8
    arrow-down
    23
    ·
    14 hours ago

    Dude, you are talking about 8 months ago. The versions you pay for are really good.

    Seriously

    • discocactus
      link
      fedilink
      arrow-up
      28
      ·
      14 hours ago

      They’re good at things you’re not good at. If you’re good at something, and you see them try to do it, they’re shite.

      • SorryQuick@lemmy.ca
        link
        fedilink
        arrow-up
        1
        ·
        1 hour ago

        LLMs are already better than humans at math and cybersecurity, since around spring this year. If you are a human in those fields, you’re worse, period. I’m not sure why you’d just to around blatantly lying about things you clearly don’t understand.

    • James R Kirk@startrek.website
      link
      fedilink
      English
      arrow-up
      10
      arrow-down
      1
      ·
      14 hours ago

      I keep hearing this but what makes them “good”? Like at the end of the day they’re still just LLMs right? Even if they can generate a perfect picture that no human could tell is AI, so what? What is that skill useful for besides fooling humans into believing something is real that isn’t true?

      Heck- they can’t even set a reminder like natural language processing (Siri) could do fifteen years ago.

      • RestrictedAccount
        link
        fedilink
        arrow-up
        2
        ·
        1 hour ago

        For instance:

        I asked Claude to help me create a website. IMHO the one it built (with some back and forth) was way better than websites I have paid professionals to build.

        Furthermore, it showed me how to get it hosted for free.

        It created programs that keep it updated that do not involve using Claude.

        I currently have a step where I am involved in the updates, but once I have confidence that it can properly handle edge cases I will have it completely automate the process.

        Siri cannot do that.

        MIT says AI can credibly do undergraduate work

        https://www.washingtonpost.com/education/2026/08/25/ai-can-now-credibly-complete-most-undergraduate-assignments-mit-warns/

        People keep piling into the copium but they should be paying attention.

        • James R Kirk@startrek.website
          link
          fedilink
          English
          arrow-up
          2
          ·
          edit-2
          57 minutes ago

          I think you’re misunderstanding my question, I’m not saying LLMs can’t spit out a website so perfectly convincing that it would be impossible to discern from a human made website.

          I’m asking what makes speeding up the creation of such “slop-sites” useful or desirable to anyone?

          EDIT: Maybe you could sell me on why automating undergraduate work is something interesting or helpful?

      • brave_lemmywinks
        link
        fedilink
        English
        arrow-up
        6
        arrow-down
        12
        ·
        13 hours ago

        Where I work it has become standard to use these tools. Being dismissive of it’s capabilities is a ship that sailed 2 years ago, and unfortunately it will only get more powerful.

        I hate it though and wish we could put all this back into Pandora’s box.

        • James R Kirk@startrek.website
          link
          fedilink
          English
          arrow-up
          11
          arrow-down
          1
          ·
          13 hours ago

          I know you think they’re good, I asked what makes them good. Maybe you can you give an example of what makes them good? It might help me understand your perspective.

          • Shindo66
            link
            fedilink
            arrow-up
            2
            ·
            12 hours ago

            Theres a really interesting episode of the podcast “diary of s ceo” with the guest Ed Zitron. He says a lot about how we can see some really good uses for AI but not trillions of dollars worth of use. Billions, maybe, but trillions? Personally, im wondering if this is a way for the uber rich to suck as much money out of the rich before the economy collapses.

            • James R Kirk@startrek.website
              link
              fedilink
              English
              arrow-up
              1
              ·
              60 minutes ago

              The investment speculation angle really does feel like crypto 2.0. Blockchain tech has some interesting, though niche right now, uses.

              I’m with Ed I absolutely don’t think LLMs are worthless but anyone who thinks a critical mass of real people want to pay real dollars for slop are living a fantasy.