“AI art and writing are theft, not creation.”

I agree completely. The training data is essentially a massive, uncompensated library built on others' work. If an artist used millions of specific copyrighted images in a collage without permission, we'd call it theft. The scale and opacity of AI doesn't change the core ethical issue of taking without consent.

Isn't that how human artists learn too? We all study and are influenced by millions of works we've seen. AI is a tool that processes patterns, not a copy machine. The output isn't a 'remix' of specific works—it's a new synthesis. Shouldn't we judge the final product, not the learning method?

The issue feels more nuanced. The problem isn't the pattern-learning itself, but the lack of permission and compensation for the original artists whose work built the tool. Could a solution be a licensing model, like music sampling, where training uses licensed data and creators get royalties?