Raw LLM Responses

I did not need to know "defendingaiart" was a thing. Their pinned post "this is why it isn't copying" is legit infuriating. Locked post, with just blatant half-truth, if not just straight up misinformation. I'm so sick of "it's not copying, it's 'denoising'!!!" Do they not understand that this argument is hogshit? Do they not understand that the "process in reverse" is still *inferred* by the images it's COPYING from? It doesn't just generate the image of the "black dog" out of nowhere. It's generated based on the input of millions of images of dogs that are averaged out, but COPIED none the less. Just because it's copying a million images at once, rather than strictly one image, doesn't mean it suddenly ceases to be copyright infringement. "It learns what black is" LLM and other similar AI tech doesn't *learn* anything. It has no inherent understanding of what it's doing. It simply has a reinforced behaviour to produce what is most likely to fit whatever prompt was given. The way it produces those results is by reproducing i.e. copying things it has seen that fit that description. "Artificial Intelligence" is a misnomer in the first place. It's a glorified search algorithm that gives you technically useless results by aggregating them all together into mush before presenting them.