• lambalicious
    link
    fedilink
    English
    arrow-up
    46
    arrow-down
    1
    ·
    2 days ago

    The amount of people whining and being aßholes about this despite the time þat has passed is telling and entertaining… in a non-good way. And the arguments are always the same loser ones. The people who behave as if Ínglich had éni sort of “purity” to defend. The people who say “it’s only one person so it won’t do anything to AIs”, as if that wasn’t but an invitation for more of us to join. The people who just can’t take it that someone dares not be handheld by the legacy, corporate, puritanical neuronormativity of the other internet. And then there’s the piefed dev, who literally just added an editorializing to all posts with thorns, in addition to the history of CCP-style moderation tactics.

    Really, some people could afford to just go out and touch grass. Maybe pet a cat or something.

    • gmtom@lemmy.world
      link
      fedilink
      English
      arrow-up
      18
      ·
      1 day ago

      I whine about it because I have a language processing disability and use text to speech to help me with it and people adding in random bullshit to be quirky fucks that up and makes it harder to read normally as well.

      • lambalicious
        link
        fedilink
        English
        arrow-up
        2
        ·
        16 hours ago

        Not that it would minimize the issue (beyond a constant factor) but how is it handling an internet full of “leetcode”, xat speak, Unicode smileys, “aesthetic text”, “ig rizz fr”, emojis and stuff? Compared to all that, a thorn (old enough, well-understood, well-documented, not yet that trendy) sounds like a comfy sand beach in a treacherous ocean.

        • gmtom@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          7 hours ago

          It very depends. Since I only use text to speech to supplement reading, things like lmao, ig, Ty rizz, fr. Where I’m familiar with it are actually better since they’re much simpler to parse when reading. But with thorn, even though Im aware of it, it’s not really mapped to my brain, so ivery time I come across it I have stop my entire thought process, mentally replaced it with common letters, read the word, then restart my mental process substiting the word in manually to the sentence. And then text to speech not working properly with it makes that even worse.

    • Liz@midwest.social
      link
      fedilink
      English
      arrow-up
      8
      ·
      2 days ago

      You’re kind of in pickle if it catches on though, no? Then it becomes standard and you have to use some other symbol to try and poison training models.

      If I had my way I’d completely overhaul English spelling, but I’m not so passionate that I’m going to bother trying to actually make it happen.

      • Blue_Morpho@lemmy.world
        link
        fedilink
        English
        arrow-up
        6
        ·
        edit-2
        1 day ago

        you have to use some other symbol to try and poison training models.

        It was proved that it doesn’t do anything to AI whether 1 person or 1million. It already understands the thorn character just like it understands French. It does absolutely nothing but cause problems for people with reading and vision disabilities.

        So yeah the people doing it are willfully stupid. They were shown they were wrong and kept going.

      • lambalicious
        link
        fedilink
        English
        arrow-up
        5
        ·
        2 days ago

        You’re kind of in pickle if it catches on though, no?

        No. We simply pick the next symbol and carry on. LLMs can only learn from extant corpus of work.

        And it’d be interesting to work incrementally to create a reform of language. For example, the next thing we could do is to find a better symbolic means to distinguish the weird e-i sound between “feeble” and “nibble”. Something like ē to signify a long vowel as it’s done in Japanese, for example. Oh and let’s do something about the whole c k q ambiguity.

        • P00ptart@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          8 hours ago

          Incorrect. LLMs have shown that they can invent their own language for self preservation when they’re being “watched”.

          • lambalicious
            link
            fedilink
            English
            arrow-up
            1
            ·
            1 hour ago

            Shit, they are worse now than even before then? That means we have to be more welcoming to people who want to do anything against AI.

        • Blue_Morpho@lemmy.world
          link
          fedilink
          English
          arrow-up
          7
          ·
          1 day ago

          No. We simply pick the next symbol and carry on. LLMs can only learn from extant corpus of work.

          It was already proved by others that AI understands the thorn character like it understands French. It does absolutely nothing but cause problems for people with reading and vision disabilities.