>>722981464
>>what AI shits out
>Why would I want any shit?
i totally agree and i am a slop artist myself. i dislike when even my gens are low quality and i also dislike soulless slop done by human hand. bad art is bad art no matter what.
so post your art anon, lets us judge you by the merits you prosthelytize.
>But, the "problems" with AI don't stop there. It's a power guzzling, IP stealing, pile of goddamn shit. It's by no stretch of the imagination producing anything. It's by technical definition and factual observation a data destroying process.
you don't fully understand what you are talking about. "ai" is not a single monolithic thing. the local model i run on my computer has nothing to do with the lake-evaporating data centers that openai runs, and it takes no more energy to generate a lewd image of mai waifu than it does to play a video game. neither does it invalidate or destroy any human made works. that's a bit of an exaggeration, don't you think?
>The more its used the more information is corrupted and loss by the process.
an ai model does not "take" pictures from the internet and stitch them together like ye olde photoshoppe. it merely knows what humans think is aesthetically pleasing, just like an LLM knows how to write a convincingly factual sentence, but not actually what any of it means or if it is even true. being "trained" on images doesn't mean it's tracing, stealing or deleting the original art, it is only observing them and making note of artistic patterns.
>The more it takes from itself the further this process happens until nothing is left but hash.
LLMs and image generators are not a monolithic singular entity that recursively studies itself. there are thousands of completely separate independent ai constructs that are all trained and used for totally different purposes.
it's true that lots of commercial web-based models like chatgpt are trained on their own outputs but that is more for convenience, not a rule of life.
