Agreed, the output of the model isn't really the issue (other than providing evidence that those works are in the dataset), the inputs are. I would also like to see groups like wikipedia get something back from this too, OpenAI abused a ton of community funded resources and expects to sell that shit back to us as GPT subs.