>>69599I must admit, I find a lot of this to be strange. It's not like low effort art is new. If we are all being honest, looking through any tag on dA will give you lots of low effort photoshops of celebrities faces on violet movie stills. They are everywhere.
Perhaps I am expressing my own bias, but I think the various ai art generations that are 'low effort' are no different. But I also think that once the novelty wears off, we've just altered one kind of low effort for another.
And once that has happened, people will gravitate back towards things that require more effort. I am old enough to remember when people claimed that photoshop was going to destroy art itself around twenty years ago, after all.
I also think that, having used plenty of ai art things, that actually getting good results requires a lot of effort. It requires knowledge of how to fine tune prompts, and what you see is likely one result out of hundreds; it's rather time consuming.
and while I respect artists who can pick up a pencil and do art, I also grasp that for someone like me, whose autism makes hand eye coordination very difficult, who will never be able to draw without extreme difficulty, the ability to make something based on my own imagination is very liberating.
I can grasp that people do not like spam, and I also don't think that people should spam art. But we already had people spamming other things, and we didn't sit here and say that the problem was photoshop existing. So let's not claim that the problem of spam is due to AI art. The problem of spam is people posting the most mundane, low effort things.
And in that regard, I think they should be encouraged to push themselves towards things that will make them stand out, not told to stop doing it entirely, as the point and desire of sharing things is because they want people to see them and engage with them. The best way to get engagement though, is to post things that are different and appealing, and low effort things rarely are.
Side note though: I do not think that dA's idea of training a model based on their own art is a good idea, simply because having trained models myself, there's far too much awful art on their site, which would skew the results, and if the art is not tagged appropriately in the model, you're basically using a smaller, worse version of Laion-B. Put it bluntly, garbage in, garbage out. There's good stuff on dA you could use for a model, but I would imagine that it's a minority of the dataset if you actually looked at it.
