zlacker

Isn't also because of OpenAI scraping the internet that companies got the walls up. How else is anyone able to gathering training data these days?

>>wesapi+(OP)
Generally speaking for a base model this isn't nearly as important as it sounds because the specifics of the data don't matter as long as there's enough of it. You may remember this from high school as the central limit theorem.

For specific things like new words and facts this does matter, but I think they're not in real trouble as long as Wikipedia stays up.