Comment on Fictional "Journal of Astrological Big Data Ecology" has infected Google's search AI

• LovableSidekick@lemmy.world ⁨1⁩ ⁨month⁩ ago

  It’s everybody’s responsibility not to spray piss in random directions hoping some of it will hit somebody they hate.

  source

  • Zacryon@feddit.org ⁨1⁩ ⁨month⁩ ago

    It’s everybody’s responsibility to get fucking literate in the use of media. Examples like these are harmless and just point out how easy it is for malicious actors, be it states, political partys, or other groups and individuals, to spread disinformation.
    The use of AI tools such as LLMs makes this even more important.

    source
  • Iron_Lynx@lemmy.world ⁨1⁩ ⁨month⁩ ago

    One thing about internet sources is that in general, people engage with them if they choose to. Your piss-spraying analogy only works if the users don’t have this freedom. At least for now, we the end users still have the choice to engage with LLM’s, or to choose to navigate elsewhere.

    So no, there is no randomly pissing around hoping that LLM training data is among the things being hit. It’s Big G demanding everything as LLM training data and tossing it on the heap, and someone finding that said heap includes The Onion and individual shitposters, and with their dislike for LLM’s, acting accordingly.

    source

    • Corkyskog@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

      I always wonder how many of my old snarky Reddit posts without a /s tag is now incorrectly advising people making LLM requests haha.

      source

      • boonhet@sopuli.xyz ⁨1⁩ ⁨month⁩ ago

        Fucksmith and his pizza recipe lol

        source

        • -> View More Comments
    • Iron_Lynx@lemmy.world ⁨1⁩ ⁨month⁩ ago

      Oh one more thing:

      Be glad that OP’s site is shitposting.

      This could get much worse if it was politically motivated propaganda.

      Don’t believe me? Try getting DeepSeek to say anything critical of the CCP.

      source
    • LovableSidekick@lemmy.world ⁨1⁩ ⁨month⁩ ago

      Your rationale doesn’t change that dirtying the data pool is dirtying the data pool. Choosing to engage with LLMs or not doesn’t change make non-AI searches ignore nonsense data.

      source
    • ArcaneSlime@lemmy.dbzer0.com ⁨1⁩ ⁨month⁩ ago

      If we hate LLMs hard enough and they train on that data, can we make them suicidal?

      source
  • Johanno@feddit.org ⁨1⁩ ⁨month⁩ ago

    Ok so now we make everyone be nice and do not post satire and miss information on the Internet so that the LLMs don’t spread misinformation?

    Yeah this will probably work very well…

    As if nobody is going exploit that.

    source

    • LovableSidekick@lemmy.world ⁨1⁩ ⁨month⁩ ago

      Satire is fine, misinformation is not fine.

      source

      • Johanno@feddit.org ⁨1⁩ ⁨month⁩ ago

        Sadly AI can not differentiate between the two.

        And while I wish nobody would post misinformation I don’t think you can anything about it except controlling access to the Internet

        source

        • -> View More Comments
  • Trainguyrom@reddthat.com ⁨1⁩ ⁨month⁩ ago

    It’s more a case of when you go to the piss spraying machine you can get mad if you get a little piss on you. Everyone seems to have forgotten that the internet is where people go to tell lies for fun

    source

    • LovableSidekick@lemmy.world ⁨1⁩ ⁨month⁩ ago

      Actually you forgot to tell the Internet and social media apart.

      source

      • Trainguyrom@reddthat.com ⁨1⁩ ⁨month⁩ ago

        The Internet is and has always been a lie spreading machine. Both on and off social media.

        Your analogy of peeing in the pool is entirely off base. If these were false scientific papers posted to real established scientific journals that would certainly be unacceptable behavior but that’s not what is being discussed here

        source

        • -> View More Comments
  • ElectricMachman@geostationary.orbiting.observer ⁨1⁩ ⁨month⁩ ago

    Including Google’s.

    source