Google researchers have come out with a new paper that warns that generative AI is ruining vast swaths of the internet with fake content.

Google researchers have come out with a new paper that warns that generative AI is ruining vast swaths of the internet with fake content — which is painfully ironic because Google has been hard at work pushing the same technology to its enormous user base.

        • @derangerEnglish
          25•12 days ago

          I disagree, you can completely avoid both FB and Twitter. You’re not going to find any valuable info on either, except maybe some cutting edge current event things on Twitter. If either of them were completely wiped tomorrow my life would be unchanged.

          Google on the other hand used to be a great resource for finding good info, but that’s ruined and getting worse as we speak. I’d argue its decline is significantly more impactful than what’s happened to FB/Twitter. It impacts me on a daily basis.

          • @[email protected]English
            1•11 days ago

            That is a valid point. Google actually provides plenty of utility while also spying on everyone. However, Meta provides a little bit of something to some individuals, but mostly just a whole bunch of nothing to most people, while still spying on everyone.

            One of the most annoying thing are the restaurants that can’t be bothered to make a real website. Instead, they just think that making a FB page is good enough. I’ve even seen some pages that ask me to log in to check the opening hours of that restaurant. Well, what if i don’t even have a FB account? I guess, I’ll just walk to that other restaurant that actually has a proper website.

  • andrew_bidlawEnglish
    33•12 days ago

    LLM is the insanely productive content creator. We can’t say how much of the web is generated by it at any moment (and that’s ignoring older copypaste articles), but the organic material one wants to prioritise in machine learning gets significantly reduced. This tech, if not isolated from it’s learning material, is predictably falling into a feedback loop, and at each cycle it is going to get worse.

    Surprisingly, pre LLM-boom datasets can probably become more valuable than contemporary ones.

    • @NazEnglish
      6•12 days ago

      I remember reading that from 2021-2023, LLMs generated more text than all humans had published combined - so arguably, actually human generated text is going to be a rarity

  • @[email protected]English
    26•12 days ago

    This person already has a vague sounding meeting on their calendar from an HR rep, their supervisor, and maybe a VP. To align vision and expectations.

  • @[email protected]English
    13•12 days ago

    Ironic.

    ““Manipulation of human likeness and falsification of evidence underlie the most prevalent tactics in real-world cases of misuse,” the researchers conclude. “Most of these were deployed with a discernible intent to influence public opinion, enable scam or fraudulent activities, or to generate profit.””

    Who could have seen that coming? But in all seriousness, this is exactly why so many people have been so vehemently opposed to generative AI. It’s not because it can’t be useful. It’s literally because of how it is actively being used.

  • @[email protected]English
    11•11 days ago

    “Google looking for a different guy to blame for search enshitifiaction when internal documents point to them ruining it long before ai was a problem”