I know it's a new technology, and we're told that it's all complicated, but it just seems fairly obvious to me that using anything in training data is creating a derivative work from it, ergo consent should be needed beyond fair use stipulations, just like using someone's text in a book, or their music in a video.

All the talk about it being so new and different just seem like deliberate obfuscation as a stalling tactic against regulation.

#GenerativeAI #LLM #AI #StochasticParrots

Because we were looking for more things to do when these clowns decided to write "the letter," and cite our #StochasticParrots paper while saying the opposite of what we write, we @emilymbender Angelina McMillan-Major and @mmitchell_ai wrote a statement in response.
https://www.dair-institute.org/blog/letter-statement-March2023