Law is a field where context is essential: to analyze a particular set of facts you need not solely the text of the law, but the full historical use and interpretation of that text in regards to other sets of facts. Speculation based on what one imagines the law to be is rarely useful in understanding law or legal arguments.
The rest of that segment is "including whether such use is of a commercial nature or is for nonprofit educational purposes;"
And then you go on to say "context is essential" while using the text of the law to justify your opinion. The "full historical use and interpretation of that text in regards to other sets of facts" has shown that original purpose has little to no bearing on trying to claim fair use.
Also, the list is not exhaustive. Those factors are considered, but they're not the entire test. Even the purposes themselves are examples, not an exhaustive list.
So why the original was created has no bearing on whether or not you've infringed copyright.
Their main thrust however seems to be that they're metaphorically making paper mache whales from shredded copies of whale books. Yes, the entire text of Moby Dick is on that whale, but it is not Moby Dick, you can't extract Moby Dick from it, you couldn't even tell Moby Dick was used in its entirety.
Which is a much better argument than "bruh, we gave it to a computer".
> It would be like saying, I'm not consuming this media for enjoyment, but to learn how to create media, therefore it's ok that I'm not compensating the creators for that media.
> That would not fly.
The copyright statute explicitly contradicts you, by requiring courts to consider the purpose and character of the infringing use. Perhaps you meant to make some other argument?
Because part of OpenAI's argument seems to be based on the intent of the creator on how the media is to be consumed. Which doesn't matter.