• altkey@lemmy.dbzer0.com
    link
    fedilink
    English
    arrow-up
    0
    ·
    10 hours ago

    Some search queues return me pages of word-to-word copypastes of one ‘original’ AI gen article. On top of existing power-googling, you need not to include words that are particularly popular in slop. It brought a need to rewrite a queue a couple of times before it gets to the point.

    They learnt on the worst examples of SEO, and then intertwined with it.

      • altkey@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        0
        ·
        9 hours ago

        It depends, that’s why I said it requires reruns with some slight modifications. When I was looking for a FOSS or at least Linux-friendly software for some live video visuals manipulation, VJ and everything with ‘video’ triggered a wave of slop, but ‘projecting software’ lead me to a rabit hole of actual list of choices, albeit most of them were paid, proprietary and Windows only.

        It’s counter to my previous expirience of including certain words to narrow the search: now I watch for what keywords bring most AI articles and drop\change them.

  • Ephera@lemmy.ml
    link
    fedilink
    English
    arrow-up
    0
    ·
    14 hours ago

    That’s always kind of bothered me about the whole generative AI thingamabob. Why are we generating things and storing them, when the ability to generate more of it is right there?

    I mean, I know in lots of cases the output is extremely flaky, so it’s not as easy to just generate the same thing again, but yeah, it still feels kind of backwards…

    • stebo@sopuli.xyz
      link
      fedilink
      arrow-up
      0
      ·
      5 hours ago

      it’s funny how my teachers used to say don’t trust everything you read on Wikipedia because anyone could edit it but now it’s among the most trustworthy websites around

    • moonlight@fedia.io
      link
      fedilink
      arrow-up
      0
      ·
      10 hours ago

      Yep, if your search engine of choice supports it, using ‘before:2022’ yields drastically better search results.

    • GoodLuckToFriends@lemmy.today
      link
      fedilink
      English
      arrow-up
      0
      ·
      15 hours ago

      Pfffft. Not far enough. You have to go back to 2015 to even find articles that aren’t shitty lists or repeating the ‘target’ word 15 times for SEO. AI has only sped up the ability to generate shit.

  • Varyk@sh.itjust.works
    link
    fedilink
    arrow-up
    0
    ·
    16 hours ago

    just went through this looking up what a daruma doll is.

    I read a couple sentences and then realized it was ai slop

  • potoo22@programming.dev
    link
    fedilink
    arrow-up
    0
    ·
    17 hours ago

    AI trains using AI generated articles on repeat. Have you seen what happens to a YouTube video when you download and reupload a video over and over again?