The Perfect Response
The Perfect Response
266 comments
-
And this is where I split with Lemmy.
There's a very fragile, fleeting war between shitty, tech bro hyped (but bankrolled) corporate AI and locally runnable, openly licensed, practical tool models without nearly as much funding. Guess which one doesn't care about breaking the law because everything is proprietary?
The "I don't care how ethical you claim to be, fuck off" attitude is going to get us stuck with the former. It's the same argument as Lemmy vs Reddit, compared to a "fuck anything like reddit, just stop using it" attitude.
What if it was just some modder trying a niche model/finetune to restore an old game, for free?
That's a rhetorical question, as I've been there: A few years ago, I used ESRGAN finetunes to help restore a game and (seperately) a TV series. Used some open databases for data. Community loved it. I suggested an update in that same community (who apparently had no idea their beloved "remaster" involved oldschool "AI"), and got banned for the mere suggestion.
So yeah, I understand AI hate, oh do I. Keep shitting on Altman an AI bros. But anyone (like this guy) who wants to bury open weights AI: you are digging your own graves.
-
Oh, so you deserve to use other people's data for free, but Musk doesn't? Fuck off with that one, buddy.
-
Using open datasets means using data people have made available publicly, for free, for any purpose. So using an AI based on that seems considerably more ethical.
-
Except gen AI didn’t exist when those people decided on their license. And besides which, it’s very difficult to specify “free to use, except in ways that undermine free access” in a license.
-
The responsibility is on the copyright holder to use a license they actually understand.
If you license your work with, say, the BSD 0 Clause, you are very explicitly giving away your right to dictate how other people use your work. Don't be angry if people do so in ways you don't like.
-
How does a model that is trained on an open dataset undermine free access? The dataset is still accessible no?
-
-
This specifically talks about AI data scrapers being an issue, and some general issues that are frankly not exclusive to open access info.
Exploitative companies are always a problem, whether it's AI or not. But someone who uses the Wikipedia text torrents as a dataset isn't doing anything of what is described in that article for example.
-
-
-
-
-
To be fair, he did say he "used some open databases for data"
-
Musk does too, if its openly licensed.
Big difference is:
- X's data crawlers don't give a shit because all their work is closed source. And they have lawyers to just smash anyone that complains.
- X intends to resell and make money off others' work. My intent is free, transformative work I don't make a penny off of, which is legally protected.
That's another thing that worries me. All this is heading in a direction that will outlaw stuff like fanfics, game mods, fan art, anything "transformative" of an original work and used noncommercially, as pretty much any digital tool can be classified as "AI" in court.
-
-
What if it was just some modder trying a niche model/finetune to restore an old game, for free?
-
Yeah? Well what if they got very similar results with traditional image processing filters? Still unethical?
-
The effect isn't the important part.
If I smash a thousand orphan skulls against a house and wet it, it'll have the same effect as a decent limewash. But people might have a problem with the sourcing of the orphan skulls.
It doesn't matter if you'we just a wittle guwy that collects the dust from the big corporate orphan skull crusher and just add a few skulls of your own, or you are the big corporate skull crusher. Both are bad people despite producing the same result as a painter that sources normal limewash made out of limestone.
-
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.
-
Dude. Read the room. You're literally in a community called "Fuck AI" and arguing for AI. Are you masochistic or just plain dumb?
-
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.
-
So it's masochism. Got it. Hey, I'm not kink-shaming. I don't get it, but you be you.
-
Yeah, maybe.
But I’m sick of getting lumped in with AI/Crypto bros for my fascination with the space. And I’m legit scared of those bros taking everything over. Guess it makes me want to shout into the void.
-
-
-
-
Even if the data is "ethically sourced," the energy consumption is still fucked.
-
Depends on what ur producing, running llama 3.1 locally on a raspberry pi doesnt produce any meaningful impact on the climate.
-
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?
-
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.
-
-
-
-
-
-
Just make a user interface for your game bro. No need to bring AI into it
-
266 comments