A human had to design the structure of the network, define its parameters, and decide what data would be used to form the network.
In a majority of the cases this simply isn’t true. Yeah, there’s some people deep into the ML game, but most predictive engines aren’t using any kind of additional fine tuning or dataset from their users. And in most stable diffusors that are popular right now, were trained on copyright violating works.
LLMs are just prediction engines, again - trained on many works that were privvy to copyright and the companies didn’t care. Unless they all can prove their dataset contains no copyright violations, which will never happen.
Image and language predictors are just that…predictors. And morally, what’s law now IS what is right. Typing some sentences into an image diffusion algorithm is no different than plugging an equation into a calculator. Math isn’t copyrightable either.
The law already has stipulations for what constitutes an AI generated work, or merely an AI assisted creation. There are clear lines drawn in the sand that most people agree with morally.