Three raccoons in a trench coat. I talk politics and furries.

https://www.youtube.com/@ragdoll_x

https://ragdollx.substack.com

https://twitter.com/x_ragdoll

  • 12 Posts
  • 40 Comments
Joined 1 year ago
cake
Cake day: June 20th, 2023

help-circle













  • This did happen a while back, with researchers finding thousands of hashes of CSAM images in LAION-2B. Still, IIRC it was something like a fraction of a fraction of 1%, and they weren’t actually available in the dataset because they had already been removed from the internet.

    You could still make AI CSAM even if you were 100% sure that none of the training images included it since that’s what these models are made for - being able to combine concepts without needing to have seen them before. If you hold the AI’s hand enough with prompt engineering, textual inversion and img2img you can get it to generate pretty much anything. That’s the power and danger of these things.


  • IIRC it was something like a fraction of a fraction of 1% that was CSAM, with the researchers identifying the images through their hashes but they weren’t actually available in the dataset because they had already been removed from the internet.

    Still, you could make AI CSAM even if you were 100% sure that none of the training images included it since that’s what these models are made for - being able to combine concepts without needing to have seen them before. If you hold the AI’s hand enough with prompt engineering, textual inversion and img2img you can get it to generate pretty much anything. That’s the power and danger of these things.







  • Ragdoll X@lemmy.worldtoGreentext@sh.itjust.worksAnon has a dream
    link
    fedilink
    arrow-up
    61
    arrow-down
    1
    ·
    edit-2
    3 months ago

    Analog greentext horror? 🤔

    also volume warning that shit’s loud


    Edit: I just thought of something. As with any 4chan story I assume this is simply fake, but putting skepticism aside for a second… I distinctly remember one day when my grandmother was on the phone and her call got incorporated into my dream because I could hear her talking. In the off-chance that they’re telling the truth I’d guess this anon just heard the call/voicemail while they were sleeping and it got incorporated into their dream.



  • Ragdoll X@lemmy.worldtoGreentext@sh.itjust.worksAnon is in love
    link
    fedilink
    arrow-up
    5
    arrow-down
    1
    ·
    edit-2
    3 months ago

    Maybe my comment came out sounding a bit too pretentious, which wasn’t what I intended… Oh well.

    To one extent or another we all convince ourselves of certain things simply because they’re emotionally convenient to us. Whether it’s that an AI loves us, or that it can speak for a loved one and relay their true feelings, or even that fairies exist.

    I must admit that when reading these accounts from people who’ve fallen in love with AIs my first reaction is amusement and some degree of contempt. But I’m really not that different from them, as I have grown incredibly emotionally attached to certain characters. I know they’re fictional and were created entirely by the mind of another person simply to fill their role in the narrative, and yet I can’t help but hold them dear to my heart.

    These LLMs are smart enough to cater to our specific instructions and desires, and were trained to give responses that please humans. So while I myself might not fall for AI, others will have different inclinations that make them more susceptible to its charm, much like how I was susceptible to the charm of certain characters.

    The experience of being fooled by fiction and our own feelings is all too human, so perhaps I shouldn’t judge them too harshly.


  • Please tell me how an AI model can distinguish between “inspiration” and plagiarism then.

    […] they just spit out something that it “thinks” is the best match for the prompt based on its training data and thus could not make this distinction in order to actively avoid plagiarism.

    I’m not entirely sure what the argument is here. Artists don’t scour the internet for any image that looks like their own drawings to avoid plagiarism, and often use photos or the artwork of others as reference, but that doesn’t mean they’re plagiarizing.

    Plagiarism is about passing off someone else’s work as your own, and image-generation models are trained with the intent to generalize - that is, being able to generate things it’s never seen before, not just copy, which is why we’re able to create an image of an astronaut riding a horse even though that’s something the model obviously would’ve never seen, and why we’re able to teach the models new concepts with methods like textual inversion or Dreambooth.




  • In early 2019 bittorrent’s website views fluctuated between ~6M to ~9M. Now it’s around 3M to 4M.

    In early 2019 utorrent’s visits fluctuated between ~26M to ~75M. Now it sits around 25M to 21M.

    The fact that there were far more captures in early 2019 for both of them might be an indication that this was their peak, and while visits have reduced since then they’re far from dying.

    Streaming services may be part of the reason, though I also think it’s because many games and software have switched to freemium & microtransactions so spending money is optional, along with the fact that free and open source alternatives to mainstream software have become more robust and popular. When I was a kid I torrented Sony Vegas, but now that’s simply not necessary since we have DaVinci Resolve.