Comment on eYou promises to change social media. But the alternatives already exist

stabby_cicada@lemmy.blahaj.zone ⁨1⁩ ⁨day⁩ ago

• doriandiaconuro@lemmy.world ⁨1⁩ ⁨day⁩ ago

  Funny timeline we live in

  source
• daychilde@lemmy.world ⁨1⁩ ⁨day⁩ ago

  The alternative is Facebook with lies that go unchecked completely. This is actually an area where AI is not bad.

  source

  • FreddyNO@lemmy.world ⁨20⁩ ⁨hours⁩ ago

    The system that is notorious for lying being used for fact checking. Yea maybe you should write “bad” in caps lock one more time, that will make you right.

    source
  • stabby_cicada@lemmy.blahaj.zone ⁨13⁩ ⁨hours⁩ ago

    “Correcting” incorrect information with more incorrect information doesn’t improve the situation.

    AI tools are inherently unreliable because of the randomness of their text generation.

    And worse, Europe doesn’t build its own AIs. LLM fact checking would have to be done by Grok or Claude or some other product from big American tech. And there’s an obvious problem with a social media network trying to avoid American censorship and political bias but “fact checking” with a tool that has American censorship and political bias built into it.

    source
  • LwL@lemmy.world ⁨16⁩ ⁨hours⁩ ago

    I doubt it, honestly. It’d likely catch a lot of misinfo, yes, but it would likely also classify any new findings that run counter to previous assumptions as misinfo. LLMs can’t keep up to date. And they still have the same issue that whoever trains them gets to decide what is and isn’t misinfo, which starts being a problem when it’s an ubiquitous social media site.

    source
  • LuceVendemiaire@lemmy.dbzer0.com ⁨1⁩ ⁨day⁩ ago

    just because it is a solution doesn’t mean its a good solution or even better than no solution

    source

    • mimavox@piefed.social ⁨1⁩ ⁨day⁩ ago

      If i’s implemented the right way, it could be. AI can be used for good things even of the knee-jerk reaction of so many people online is to equate it with crap.

      source

      • LuceVendemiaire@lemmy.dbzer0.com ⁨1⁩ ⁨day⁩ ago

        Recoiling upon smelling shit is also a kneejerk reaction

        its always this same bullshit, “if we just implemented this correctly” where can an AI participate in fact-checking? It can’t be trusted because of hallucinations, so the solution would be to uh… manually review everything it does? just rely on third parties to do it? What ACTUAL USE does this shit have?

        source
      • Nalivai@lemmy.world ⁨1⁩ ⁨day⁩ ago

        It couldn’t be. Lying bias machine that gives people psychosis can’t magically stop being what it is. So it will always be terrible and unnecessary at best, harmful at regular.

        source
  • Specter@feddit.org ⁨1⁩ ⁨day⁩ ago

    I mean, there’s always Lemmy.

    source