> there's clearly no copyright infringement in a human learning from someone's work and creating their own output, even if it "copies" an artist's style or draws inspiration from someone's plot-line.

What do you mean here by "clearly?" This is not at all clear, and court cases have been decided in the opposite direction.

This case: https://www.reuters.com/article/lifestyle/marvin-gaye-family...

is as far from what you say is "clearly" true as could possibly be. You're handwaving away the parts of the question that are difficult.

But that case involves publishing of content. If you want to compare them, it seems like you'd argue that the content of Pharrell Williams' head is somehow copyright infringement. I've yet to see anyone credible say that an AI model outputting exact content and then someone publishing that should be allowed. If you manage to make an AI model output copyrighted content, you can't then claim it's not copyrighted. If you sit on one side of a table and read a book out loud and I write it down (with minor transcription errors), that content is almost certainly still copyrighted content and I could not distribute it legally.

In the case you linked, it wasn't infringement for Robin Thicke to be impressed by Marvin Gaye. It was infringement to create a substantial copy.

Applying this same legal doctrine to LLMs, it's totally fine for them to train on copyrighted works. A problem would only arise if they produced substantial copies of those works.

It may be difficult to quantify the harm done in that case. Robin Thicke sold millions of copies of blurred lines. Each LLM output is generally consumed by a single person. There could be millions of examples of blatant copyright infringement by LLMs that will never come to light in court. There also could be vanishingly few. We may need new regulations to police it.

. If you analogize to human leaning

You can't make valid legal analogies to human learning when dealing with copyright law, because human brains are not a fixed media under copyright law, thus impressions in human brains are not copies of any kind under copyright law, thus "well, when you make an impression in the human brain, its not a copyright violation" is never a good legal analogy for when you make something that is in a form which can be a copy under copyright law.

> For AI training, a central argument is that training is transformative

This doesn't actually matter though does it? They still had access to a copy of the data in the first place to train the AI on

Since they likely did not pay a license to have access to the books they trained the AI on, then they violated copyright

The same way it would be violating copyright for a university student to pirate a textbook and learn from it

It matters because that is what the plaintiffs are arguing - that the use of copyright material for AI training is not fair use. They are not solely arguing that meta engaged in copyright infringement through piracy.

One could argue a student going to a library and reading a book is analogous and legal.

Information contained in subjective mental experience is not a medium which can qualify as a copy (infringing or otherwise) in copyright law, whereas data recorded in digital media such as computer memory is, so they are not similar circumstances with regard to copyright law, however analogous you might feel they are from some other perspective.

What if there was a camera watching the student read in the library, such that the words could be seen?

That doesn't change anything about the student reading.

The recording made by the camera would be a whole different issue.

> whereas data recorded in digital media such as computer memory is

Exactly.

Plus, Thomson Reuters recently won a case regarding using copyrighted material for AI training:

https://www.wired.com/story/thomson-reuters-ai-copyright-law...

> Thomson Reuters has won the first major AI copyright case in the United States. In 2020, the media and technology conglomerate filed an unprecedented AI copyright lawsuit against the legal AI startup Ross Intelligence. In the complaint, Thomson Reuters claimed the AI firm reproduced materials from its legal research firm Westlaw. Today, a judge ruled in Thomson Reuters’ favor, finding that the company’s copyright was indeed infringed by Ross Intelligence’s actions.

Though a cynical way to read this would be that the larger corporation won against the smaller, so the actual precedent here is Meta winning.

But as others have noted in these comments, even that has a silver lining: should Meta win, then everyone can pirate books to read, right? /s

If I remember right, that particular case was Ross Intelligence almost copying a database of Westlaw's summaries completely and with the intent to compete directly too, with their only real addition being adding AI to the search capabilities, making it much more nuanced of a decision than just ruling against the use of AI.

Thank you for confirming what I meant. The crux is learning is learning.

Pirating the book is copyright infringement ... reading in a library is not.

Training of a neural network on copyrighted work is the algorithmic equivalent of a "subjective mental experience".

> The same way it would be violating copyright for a university student to pirate a textbook and learn from it

Presumably, there would be no requirement for the student to learn from it - making the unauthorised copy would be itself an infringement.