I notice a large number of ragebait-y political communities being spun up by new users with thousands of posts & ai profile header photos. I notice comment sections are more acrimonious, and foreign disinfo talking points are circulating a lot more prolifically than before the US election started ramping up.

Anyone else notice this? Any idea on how to combat it on this platform? Are there any communities built around creating block lists of obvious troll/ai/disinfo accounts & communities?

  • fury@lemmy.world
    4·
    3 months ago

    I wish they worked on pictures too :( half my feed is still full of stuff I’ve blocked just because they didn’t use the word in the title

    • Feathercrown@lemmy.worldEnglish
      1·
      3 months ago

      OCR is easy enough now that I’m a little surprised text isn’t auto-parsed from uploaded images

      • talEnglish
        3·
        3 months ago

        Hmm.

        Having a bot that auto-OCRs images and comments with the text in the image – and, even better, a AI-generated description, maybe – would probably help vision-impaired users.

        I mean, it’d be technically more-efficient to have a bot run it as a one-off than to have every impaired-vision user run it.

        I was actually just looking at a software package that did something like that (for the purpose of removing text from AI-generated images…it’d run text recognition on the image, identify text areas, then automatically inpaint the identified region, repeating until the text was gone).

        https://github.com/iuliaturc/detextify

        So I know that the underlying software is out there.