I still think it's a pretty clear conceptual mistake to view using a piece of content in training an AI model as "stealing" it or "copying it without permission." And the insistence by many on doubling down on that conceptual mistake makes conversations about the ethics of AI less fruitful.
