• frunch@lemmy.world
        link
        fedilink
        English
        arrow-up
        6
        ·
        2 months ago

        But they’ll have stragglers just the same as any major social media site. Even though many here have standards they won’t easily abandon, there are scores of people that won’t even know if/when AI started being used on the site or would care enough to leave if they did.

        Plus every time we leave a platform we need to find or build a new one. The time it takes to get others to migrate and develop into a worthwhile community is hard to predict and it may not even work out. It sucks social media is such shit anymore, but it seems inevitable that it will remain that way given the landscape of the Internet at this point.

        I say this as someone who’s drifted from Fark to Digg to Reddit to Lemmy over the past 20-25 years 🎈 (zero loyalty as well)

  • SkaveRat@discuss.tchncs.de
    link
    fedilink
    English
    arrow-up
    58
    arrow-down
    2
    ·
    2 months ago

    “Well, WE won’t train on your data. But this subsidiary company we created on the other hand…”

        • Woovie@lemmy.world
          link
          fedilink
          English
          arrow-up
          6
          arrow-down
          1
          ·
          2 months ago

          okay but they use ai in production for moderation easing and detection, so you didn’t even bother to read why they said this. this whole site is full of fucking stupid presumptuous comments like this because it’s easier to blindly rage.

  • AbouBenAdhem@lemmy.world
    link
    fedilink
    English
    arrow-up
    28
    ·
    edit-2
    2 months ago

    If the AT protocol allows public access to content, they can’t create a proprietary training set. But the content is available for anyone who wants to add it to a public training set.

  • LegoBrickOnFire@lemmy.world
    link
    fedilink
    English
    arrow-up
    23
    ·
    edit-2
    2 months ago

    To be fair, “they” could probably train AI on Lemmy data, they just won’t ask for permission and won’t be charged for it

      • Excrubulent@slrpnk.net
        link
        fedilink
        English
        arrow-up
        6
        ·
        2 months ago

        Yeah, I looked into it and the backend is proprietary, so the central owner can restrict features. Like for instance independent instances can only have 10 users.

        It’s “decentralised” except only in extremely limited scope, the code is centrally controlled and the network remains largely, functionally centralised.

        They’re capitalising on the decentralised, federated buzz while doing it so poorly they’re setting up users to say “oh people tried decentralisation, it doesn’t work, look at Bluesky”.

        If it’s not open source, it’s not decentralised.

  • GiddyGap@lemm.ee
    link
    fedilink
    English
    arrow-up
    21
    ·
    2 months ago

    Bluesky is VC backed. They’ll want to make money down the road, and they’ll definitely train AI soon if not already.