OpenAI doesn't just get to steal work and then say "sorry, not possible" and shrug it off.
The NYTimes should be suing.
Copyright law is a prehistoric and corrupt system that has been about protecting the profit margins of Disney and Warner Bros rather than protecting real art and science for living memory. Unless copy/paste superhero movies are your definition of art I suppose.
Unfortunately it seems like judges and the general public are so clueless as to how this technology works it might get regulated into the ground by uneducated people before it ever has a chance to take off. All so we can protect endless listicle factories. What a shame.
You are correct, if I were to steal something, surely I can be made to give it back to you. However, if I haven't actually stolen it, there is nothing for me to return.
By analogy, if OpenAI copied data from the NYT, they should be able to at least provide a reference. But if they don't actually have a proper copy of it, they cannot.
This kind of mentality would have stopped the internet from existing. After all, it has been an absolute copyright nightmare, has it not?
If that's what copyright does then we are better without it.
When told it is impossible they go "Geek Harder then Nerd" like demanding it will make it happen.
These types of arguments miss the mark entirely imho. First and foremost, not every instance of copyrighted creation involves a giant corporation. Second, what you are arguing against is the unfair leverage corporations have when negotiating a deal with a rising artist.