• SaucySnake@lemmy.world
      link
      fedilink
      arrow-up
      48
      ·
      4 months ago

      https://arxiv.org/abs/2306.07899 here’s a paper that found that one of the biggest sources for LLM training data is corrupted by people using AI to complete the tasks. There are plenty of papers out there that show the effects of this, which they call “model collapse”.

    • FMT99@lemmy.world
      link
      fedilink
      arrow-up
      16
      ·
      4 months ago

      Same. I keep hearing folks mention this but it’s not like AI developers aren’t aware of this (apart from a bunch of shitty startups that would fail no matter what) One way to deal with it for example is Microsoft is shelling out so much for “pre-AI” datasets (Reddit) but I’m sure there’s a lot more of those kinds of initiatives.

      Google on the other hand is going to be hard pressed to deal with the ever increasing deluge of AI spam.

      • Ultraviolet@lemmy.world
        link
        fedilink
        English
        arrow-up
        26
        ·
        4 months ago

        That’s a way to deal with it, but in the long term, “pre-AI” becomes a longer and longer time ago, and less and less useful for any practical purposes.

      • I Cast Fist@programming.dev
        link
        fedilink
        English
        arrow-up
        1
        ·
        4 months ago

        Google on the other hand is going to be hard pressed to deal with the ever increasing deluge of AI spam.

        Given how they’re one of the main culprits of showing AI spam on the first page, I don’t think they care at all

    • hypertown@ani.socialOP
      link
      fedilink
      arrow-up
      6
      arrow-down
      1
      ·
      4 months ago

      While I don’t have definitive proof I’ve seen both AI fanarts that are so good it’s hard to tell if it’s AI or not and AI abominations that are so bad and artificial it want to make you puke so I guess it really depends on the model.

      Funny enough adobe now offers AI generated stock photos that are closer to those abominations rather than anything good. Though if you think about it AI stock art is so pointless… There are already so many stock photos you can choose from. Why would you go out of your way to choose a photo that looks almost the same as regular stock photos but people have 4 arms in it…

      • I Cast Fist@programming.dev
        link
        fedilink
        English
        arrow-up
        2
        ·
        4 months ago

        Why would you go out of your way to choose a photo that looks almost the same as regular stock photos but people have 4 arms in it…

        You never know what smut some people are writing. Maybe that 4 armed horror was exactly what they needed for their cover.

  • zcd@lemmy.ca
    link
    fedilink
    arrow-up
    33
    arrow-down
    1
    ·
    edit-2
    4 months ago

    AI “human-centipeding” itself with art and word salads Is going to make confirmed human generated content an extremely valuable resource. That’s partly why literally everybody is foaming at the mouth trying to scrape all of our shit. The signal (human) to noise is only going to get worse, The dead Internet is almost here if it isn’t already

    • brucethemoose@lemmy.world
      link
      fedilink
      arrow-up
      5
      ·
      4 months ago

      Its far worse because LLMs are so data hungry. Getting quality data for image diffusion models is not nearly as much of an issue, though still a problem.

        • brucethemoose@lemmy.world
          link
          fedilink
          arrow-up
          4
          ·
          edit-2
          4 months ago

          Nah. I hate to sound bitter/salty, but all the AI haters are just going to fuel OpenAI’s crusade/lobbying against open source, and we will be stuck with expensive, inefficient, dumb corporate API models trained on copyrighted material in secret because the corporations literally don’t care. And it will do nothing to solve the environmental problems.

          There’s tons of research on making training and especially inference more power efficient, on making data cleaner and fairer, and it’s getting squandered from the lobbying against open source that the “AI is all bad” crowd is fueling. All the money to even turn these experiments into usable models is getting funneled away already.

          Everyone’s got it wrong, the battle isn’t between AI or no AI, it’s whether your own it and run it yourself, or big tech owns it and runs it. You know, like Lemmy vs Reddit.

          So… that’s my rant.

  • Naz@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    7
    arrow-down
    5
    ·
    4 months ago

    Hello, I coined the term “hallucinations” way back in 2019.

    I heard “inbreeding” from another author and I’ve been using that in formal circles and I think it’s hilarious.

    Yes, we’re inbreeding the visual models ;)))))

  • gmtom@lemmy.world
    link
    fedilink
    arrow-up
    3
    arrow-down
    1
    ·
    4 months ago

    This is a problem in AI but these people MASSIVELY over state how big of a problem it is because, surprise surprise, the militant anti-AI crows are actually incredibly uneducated on the topic and just parrot the random bullshit they see online because it agrees with their existing beliefs.