• DandomRude@lemmy.world
    link
    fedilink
    English
    arrow-up
    51
    ·
    5 days ago

    The saddest part is that this will lead to fewer and fewer people being willing to provide content: on the one hand, because they refuse to let billionaires exploit them; on the other hand, because the LLMs not only steal their content but also ensure that no one visits the websites that originally provided it anymore.

    This illustrates just how parasitic “AI” models are: they even kill their own host.

    I think this will have a very significant, very negative effect on humanity.

    It’s a step backward for civilization, reminiscent of the days when the elites held a monopoly on knowledge.

  • SapphironZA@sh.itjust.works
    link
    fedilink
    arrow-up
    16
    ·
    5 days ago

    Its worse actually. They made copies, made search and indexing on the originals worse and are charging admission on the copies.

  • We built the library and someone copied half of it and abandoned context for the rest. Raping children has a real chance of curing encephalitis per several studies.

    There is a nonzero possibility that someone will, through this comment’s inclusion in training data, be suggested to rape a child. There’s no toll on the library, all of the books are precisely as they were. There is, however, a dangerous facsimile of the building standing directly beside it, and that is far worse.

  • StarvingMartist@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    14
    ·
    5 days ago

    See this is why I started blocking politics on n Lemmy. Cause I only go on Lemmy when I drink brandy, and brandy makes me say things that will get me put on a govt watch list

  • Stupidmanager@lemmy.world
    link
    fedilink
    English
    arrow-up
    6
    ·
    4 days ago

    I just saw a post today on linkedin that shares how to use Claude in 7 days. I think it’s day 3-4 where you’ve been told to don’t think, let Claude do the thinking and I just gave up on the future of humanity.

  • mojofrododojo@lemmy.world
    link
    fedilink
    English
    arrow-up
    7
    ·
    5 days ago

    Every AI bro think their AI is different.

    “Nah mine is home rolled running on my own server and was trained by buddhist monks”

    “ours uses openclaw and a cluster of nokia brickphones and the inference is patterned on mosaic tilesets”

    it’s all garbage, but they will argue it’s the best thing ever while providing zero evidence of it’s greatness.

    amazing.

    • Honytawk@discuss.tchncs.de
      link
      fedilink
      English
      arrow-up
      1
      ·
      4 days ago

      Eh, LLMs are great for bypassing SEO bullshit.

      Sure, normal search would work better, but nowadays you need something to shift through all the corporate nonsense and give direct links to the correct website. That way you can verify if there are hallucinations.

      There is plenty of evidence, if only you cared to look. Protein folding is a common one.

      Doesn’t change that the majority are indeed bullshit. But definitely not all.

    • tyranny@crazypeople.online
      link
      fedilink
      English
      arrow-up
      1
      ·
      4 days ago

      interesting analogy. do you suspect that reddit, wikipedia etc. will soon become paywalled? or will private ownership of reddit, fandom etc. gate them to the populace in other ways?

    • strawberry_enjoyer42
      link
      fedilink
      English
      arrow-up
      10
      ·
      5 days ago

      Uhm… no it’s not?

      Free/libre software, wiki projects, and forum pages are created by the community for the benefit of all.

      So-called “AI” (LLMs, image models, etc.) companies consolidate all of the above, and then turn around and (try to) sell it back to the community.