• brucethemoose@lemmy.world
    link
    fedilink
    arrow-up
    7
    arrow-down
    17
    ·
    edit-2
    2 days ago

    Even if all involved data is explicity public domain?

    What if it’s not public data at all? Like artifical collections of pixels used to train some early upscaling models?

    That’s what I was getting: some upscaling models are really old, used in standard production tools under the hood, and completely legally licensed. Where do you draw the line between ‘bad’ and ‘good’ AI?

    Also I don’t get the analogy. I’m contributing nothing to big, enshittified models by doing hobbyist work, if anything it poisons them by making public data “inbred” if they want to crawl whatever gets posted.

      • brucethemoose@lemmy.world
        link
        fedilink
        arrow-up
        1
        arrow-down
        1
        ·
        edit-2
        1 day ago

        I’m trying to make the distinction between local models and corporate AI.

        I think what people really hate is enshittification. They hate the shitty capitalism of unethical, inefficient, crappy, hype and buzzword-laden AI that’s shoved down everyone’s throats. They hate how giant companies are stealing from everyone with no repercussions to prop up their toxic systems, and I do too. It doesn’t have to be that way, but it will be if the “fuck AI” attitude like the one on that website is the prevalent one.

      • untakenusername@sh.itjust.works
        link
        fedilink
        arrow-up
        1
        arrow-down
        1
        ·
        12 hours ago

        Depends on what ur producing, running llama 3.1 locally on a raspberry pi doesnt produce any meaningful impact on the climate.

      • hedgehog@ttrpg.network
        link
        fedilink
        arrow-up
        4
        arrow-down
        8
        ·
        2 days ago

        The energy consumption of a single AI exchange is roughly on par with a single Google search back in 2009. Source. Was using Google search in 2009 unethical?

      • brucethemoose@lemmy.world
        link
        fedilink
        arrow-up
        7
        arrow-down
        17
        ·
        edit-2
        2 days ago

        Total nonsense. ESRGAN was trained on potatoes, tons of research models are. I fintune models on my desktop for nickels of electricity; it never touches a cloud datacenter.

        At the high end, if you look past bullshiters like Altman, models are dirt cheap to run and getting cheaper. If Bitnet takes off (and a 2B model was just released days ago), inference energy consumption will be basically free and on-device, like video encoding/decoding is now.

        Again, I emphasize, its corporate bullshit giving everything a bad name.