• 12 Posts
  • 175 Comments
Joined 2 years ago
cake
Cake day: July 22nd, 2023

help-circle





  • That’s an interesting perspective actually

    Maybe it’s because of who’s giving them? If my little cousin gave me an AI Christmas card, I’d be happier than if a stranger gave me one on the street. (Though I’d feel bummed if they didn’t even marker in a single custom sentence)

    i.e. higher standards of creativity/effort from a stranger than from a family member.

    Also the stranger isn’t stuffing a tenner in the card lmao




  • Ty for feedback :>

    Your paragraph read well. I definitely agree – grammar with risks, outside of hyper-formal sitches, is just stylized diction. ChatGPT could scarcely come up with an e.e. cummings poem (just tested now, it never gets the style about right), nor dare to abuse parentheses, nor remove cruft for conciseness (e.g. to start a sentence with “Kind of changed” instead of “This kind of changes” for compression (woot)). It’s a “wrong” but not quite “wrong”, and I’m glad that “riskless” manages to carry that feeling

    And I edit a lot too :) it’s the “post-email-send clarity” effect


  • edit: updated accordingly for clarity

    Ah, I mean proper grammar as in formal, largely riskless grammar. For example, AI wouldn’t connect

    monolingual + educated + have access to technology

    with pluses, like a human would.

    Not sure how I’d phrase that though. Maybe “perfect, risklessly formal grammar” as I just tried to call it? (i.e. if AI trainers consider using +‘es a “risk”, as opposed to staying formal and spick n’ span clean).

    Perfect grammar is humanly possible but there is some scrutiny that can be applied to GPT-style grammar, especially in the context of the casually-toned web (where 100%ed grammar isn’t strictly necessary!).





  • I’m not sure I agree but I’m happy to discuss! :)

    Why are you calling my statement “selective memory” (am I intentionally excluding something?), and what do you mean by “way worse”? Do you consider unskilled art as not art at all (i.e. “so-called”)?

    What I was trying to say, is that on social media, skilled artists formerly dominated attention (likes, upvotes) because viewers wanted well-constructed, pleasing-to-the-eye artwork. I wasn’t trying to say that they were the only art posters (sorry for my wording!). Continuing, now that AI is in the arena, “technically-decent” art is no longer the lower bound for pleasurable-to-see – now, viewers are more partial to knowing that a human was vulnerable when they expressed themselves with art.

    It’s an intensification of internet-ugly aesthetic, which Douglas (2014) called "an imposition of messy humanity upon an online world of smooth gradients, blemish correcting Photoshop, and AutoCorrect” (p. 314). Now, online, handmaking art at all is a declaration of humanity, because you could corporately fake something full-colored and intricate, but arguably soulless, with lower effort.

    Of course, I’ll try to take it from your perspective. I’ve seen really bad human art (I like art!), and I’ve seen less-artifacted AI art (have you ever seen Even_Adder’s generations on lemmy.dbzer0? they don’t have the overshading issue at all). Of course, some may disagree that the latter is art (is art only human expression?), but supposing I do consider the latter art, my point still stands – viewers are more on the lookout for genuineness now.

    Happy to see what you think!

    References


  • fool@programming.devtoLinux@lemmy.mlMerry Christmas, Linux Community!
    link
    fedilink
    arrow-up
    12
    arrow-down
    5
    ·
    edit-2
    4 months ago

    edit: Please be nice to each other! :(

    Lots of downvotes in this reply chain. Not to be a “I don’t wanna be either side” kinda guy but AI isn’t all bad and isn’t all good either. (Greys!)

    Merry Christmasing should be a genuine hug. Even if this was made by a homegrown open-weight open-dataset inference model, it’s nearly 100% low-effort generated – holidays need the human aspect, no? Covering yourself up too much in AI takes away from the humanness with corporate diction, and people need evidence of risktaking genuineness nowadays.

    On the other hand, AI is definitely useful… but elsewhere. It’s not strictly anti-human even if conglomerates are using it that way, which I think you agree on. Wading through HOA using local NLP setups is human. Looking through a Mandarin thread when typical translation sucks, is human.

    But there are domains for its use and there is ethical stuff to work on. This post just doesn’t fit the domain too well, as others agree…




  • I love all the pleasantly deep answers in this thread.

    For my input: I’m everything I ever was, all at once.

    You know how lenses refract over each other at the optometrist? Or how colors combine when you stack transparent cups in the washer? That’s me. I have parts from everyone I ever met, and parts from everyone I ever was. There’s no mask, even if I focus on one part of the mosaic in a meeting vs. another when I nerd out w/ a buddy – it’s all equally me.

    I’m not Shrek though. Onions have layers, but I’m prismatic glass, chips and dips and all.


  • I disagree with this sentiment; I’m inclined to believe that AI has actually lowered the bar for meaning.

    Before AI, typically only skilled artists drew pictures for the web. But now that AI is making art that’s less meaningful than crayon pictures, there’s the growing sentiment of

    I’d rather see a crayon picture than AI slop.

    which could actually mean more people have the ability to go on and artify.

    Of course this is anecdotal; it’s the reason I started drawing again :)