Asklemmy

49762 readers

447 users here now

A loosely moderated place to ask open-ended questions

Search asklemmy 🔍

If your post meets the following criteria, it's welcome here!

Open-ended question
Not offensive: at this point, we do not have the bandwidth to moderate overtly political discussions. Assume best intent and be excellent to each other.
Not regarding using or support for Lemmy: context, see the list of support communities and tools for finding communities below
Not ad nauseam inducing: please make sure it is a question that would be new to most members
An actual topic of discussion

Looking for support?

Looking for a community?

Lemmyverse: community search
sub.rehab: maps old subreddits to fediverse options, marks official as such
!lemmy411@lemmy.ca: a community for finding communities

~Icon~ ~by~ ~@Double_A@discuss.tchncs.de~

founded 6 years ago

MODERATORS

292

OK, not to be runde or anything, but why is your banner AI generated (lemmy.ml)

submitted 6 days ago* (last edited 6 days ago) by BennyTheExplorer@lemmy.world to c/asklemmy@lemmy.ml

224 comments fedilink hide all child comments

In my opinion, AI just feels like the logical next step for capitalist exploitation and destruction of culture. Generative AI is (in most cases) just a fancy way for cooperations to steal art on a scale, that hasn't been possible before. And then they use AI to fill the internet with slop and misinformation and actual artists are getting fired from their jobs, because the company replaces them with an AI, that was trained on their original art. Because of these reasons and some others, it just feels wrong to me, to be using AI in such a manner, when this community should be about inclusion and kindness. Wouldn't it be much cooler, if we commissioned an actual artist for the banner or find a nice existing artwork (where the licence fits, of course)? I would love to hear your thoughts!

you are viewing a single comment's thread
view the rest of the comments

[–] FauxLiving@lemmy.world -3 points 5 days ago (2 children)

And due to the scale of these projects and the amount of existing work they require in their construction, there are no non-exploitative GenAI systems

That hasn't been true for years now.

AI training techniques have rapidly improved to the point where they allow people to train completely new diffusion models from scratch with a few thousand images on consumer hardware.

In addition, and due to these training advancements, some commercial providers have trained larger models using artwork specifically licensed to train generative models. Adobe Firefly, for example.

It isn't the case, and hasn't been for years, that you can simply say that any generative work is built on """stolen""" work.

Unless you know what model the person used, it's just ignorance to accuse them of using "exploitative" generative AI.

[–] patatas@sh.itjust.works 8 points 5 days ago* (last edited 5 days ago)

Can you provide a few real-life examples of images made with a model trained on just "a few thousand images on consumer hardware", along with stats on how many images, where those images were from, and the computing hardware & power expended (including in the making of the training program)? Because I flat out do not believe that one of those was capable of producing the banner image in question.

[–] ClamDrinker@lemmy.world 1 points 5 days ago

You are probably confusing fine tuning with training. You can fine tune an existing model to produce more output in line with sample images, essentially embedding a default "style" into every thing it produces afterwards (Eg. LoRAs). That can be done with such a small image size, but it still requires the full model that was trained on likely billions of images.