Major hosts already support image links, and it’d be a good way to stick it to reddit et al. Especially if it looks like a captcha text. Or is it a matter of ‘aint nobody got time for that’…?

  • DerisionConsulting@lemmy.ca
    link
    fedilink
    English
    arrow-up
    21
    ·
    7 months ago
    • The data for those on mobile devices without unlimited data.
    • users with sight issues using text to speech
    • People who are using translators

    and probably a dozen more things

    • nave@lemmy.ca
      link
      fedilink
      arrow-up
      4
      ·
      7 months ago

      I also recently saw an article about YouTube videos being transcribed for GPT-4 training data so if the companies really want to they’ll get the data anyways.

    • cbarrick@lemmy.world
      link
      fedilink
      English
      arrow-up
      6
      ·
      7 months ago

      Which is like the easiest ML problem.

      Students commonly implement OCR as a homework assignment in undergrad machine learning courses (see the MNIST dataset).

  • DeepThought42@lemmy.world
    link
    fedilink
    English
    arrow-up
    10
    ·
    7 months ago
    1. It would be a pain in the ass
    2. You’d still have to post (alt)text for accessibility reasons
    3. OCR exists
  • Admiral Patrick@lemmy.world
    link
    fedilink
    arrow-up
    9
    ·
    7 months ago

    Because that would greatly harm accessibility for the visually impaired.

    And if you wanted to adhere to accessibility guidelines for the conversational images, you’d need to add alt text to them. But then the ML models can train on that, so you’re back to square one.

  • rufus@discuss.tchncs.de
    link
    fedilink
    arrow-up
    7
    ·
    edit-2
    7 months ago

    I hate people doing that here. It’s always unreadable since people took the screenshot on a PC and the text is tiny and I need to scroll sideways every 8 words… Or I’m at the PC and the screenshot is just 3 sentences without a link to read the rest.

    I’d say let them train their AI on my comments. I just want a share of the finished product in exchange.

  • Willy@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    1
    ·
    7 months ago

    most people don’t care if their comment help train ai. not sure why lemmy generally seems to.

  • NeoNachtwaechter@lemmy.world
    link
    fedilink
    arrow-up
    2
    arrow-down
    1
    ·
    7 months ago

    Your idea is that AI cannot understand the meaning of your comments. But that will ‘hold’ only a year or two. Then the hungry AI’s (if we don’t stop them in time) will incorporate character recognition from images in it’s learning.

  • Corroded@leminal.space
    link
    fedilink
    English
    arrow-up
    1
    arrow-down
    3
    ·
    7 months ago

    I think it’s better to use photos in combination with text for the reasons other people have already mentioned.

    I think you can upload images to Reddit directly now and have them automatically displayed but I feel like the process is still a tad easier on Lemmy and you see it more often here. Using AI images as examples or to add flair to a post or comment might be a better way to stick it to Reddit.