In addition to the possible business threat, forcing OpenAI to identify its use of copyrighted data would expose the company to potential lawsuits. Generative AI systems like ChatGPT and DALL-E are trained using large amounts of data scraped from the web, much of it copyright protected. When companies disclose these data sources it leaves them open to legal challenges. OpenAI rival Stability AI, for example, is currently being sued by stock image maker Getty Images for using its copyrighted data to train its AI image generator.
Aaaaaand there it is. They don’t want to admit how much copyrighted materials they’ve been using.
If I do a book report based on a book that I picked up from the library, am I violating copyright? If I write a movie review for a newspaper that tells the plot of the film, am I violating copyright? Now, if the information that they have used is locked behind paywalls and obtained illegally, then sure, fire ze missiles, but if it is readily accessible and not being reprinted wholesale by the AI, then it doesn’t seem that different from any of the other millions of ways we use data in every day life. Just because a machine learned it instead of a human, I don’t believe that it makes it inherently wrong. I am very open to discussion on this, and if anyone has a counter-argument, I’d love to hear it, because this is a new field of technology that we should all talk about and learn to understand better.
Edit: I asked GPT-4 what it thought about this, and here is what it said:
@chemical_cutthroat
The first conceptual mistake in this analogy is assuming the LLM entity is “writing”. A person or a sentient being writing is still showing signs of intellectual work, which is how the example book report and movie review will not be accused of plagiarism, which is very very basically stealing someone’s output but one that is not made legally ownership of (which then brings it to copyright infringement territory).
LLMs are producing text based on statistical probability meaning it is quite literally aping/replicating the aesthetic form of a known genre of textual output, which in these cases are given the legal status of intellectual property. So yes, an LLM-generated textual output that is in the form of a book report or movie review looks the way it does by copying with no creative intent previous works of the genre. It’s the same way YouTube video essays get taken down if it’s just a collection of movie clips that might sound like a full dialogue. Of course in that example yt clip, if you can argue it’s a creative output where an artist is forming a new piece out of a collage of previous media, the rights owner to those movie clips might lose their claim to the said video. You can’t make that defence with OpenAI.
@stopthatgirl7
LLMs are not book reports. They are not synthesizing information. They’re just pulling words based on probability distributions.Those probability distributions are based entirely on what training data has been fed into them.
You can see what this really means in action when you call on them to spit out paragraphs on topics they haven’t ingested enough sources for. Their distributions are sparse, and they’ll spit out entire chunks of text that are pulled directly from those sources, without citation.
If you write a book report that just reprinted significant swaths of the book, that would be plaigerism, and yes, would 100% be called copyright infringement.
Importantly, though, the copyright infringement for these models does not come at the point where it spits out passages from a copyrighted work. It occurs at the point where the work is copied and used for purposes that fall outside what the work is licensed for. And most people have not licensed their words for billion dollar companies to use them in for-profit products.
@Kichae
The exact same thing a human does when writing a sentence. I’m starting to think that the backlash against AI is simply because it’s showing us what simple machines we humans are as far as thinking and creativity goes.
Do you have an example of this? I’ve used GPT extensively for a while now, and I’ve never had it do that. If it gives me a chunk of data directly from a source, it always lists the source for me. However, I may not be digging deep enough into things it doesn’t understand. If we have a repeatable case of this, I’d love to see it so I can better understand it.
This is the meat and potatoes of it. When a work is made public, be it a book, movie, song, physical or digital, it is placed in the public domain and can be freely consumed by the public, and it then becomes part of our own particular data set. However, the public, up until a year ago, wasn’t capable of doing what an AI does on such a large scale and with such ease of use. The problem isn’t that it’s using copyright material to create. Humans do that all the time, we just call it an “homage” or “parody” or “style”. An AI can do it much better, much more accurately, and much more quickly, though. That’s the rub, and I’m fine with updating the laws based on evolving technology, but let’s call a spade a spade. AI isn’t doing anything that humans haven’t been doing for as long as their has been verbal storytelling. The difference is that AI is so much better at it than we are, and we need to decide if we should adjust what we allow our own works to be used for. If we do, though, it must effect the AI in the same way that it does the human, otherwise this debate will never end. If we hamstring the data that an AI can learn from, a human must have the same handicap.
It is an area that will require us to think carefully of the ethics of the situation. Humans create works for humans. Has this really changed? Now consumption happens through a machine learning interface. I agree with your reasoning, but we have an elephant in the room that this line of reasoning does not address.
When we ask the AI system to generate content in someone else’s style or when the AI distorts someone’s view in its responses. It is in this area where things get very murky for me. Can I get an AI to eventually write another book in Terry Pratchett’s style? Would his estate be entitled to some form of compensation? And that is an easier one compared to living authors or writers. We already see the way image generating AI programs copy artists. Now we are getting the same for language and more.
It will certainly be an interesting space to follow in the next few years as we develop new ethics around this.
@mack123
No, that’s fair use under parody. Weird Al isn’t compensating other artists for parody, so the creators of OpenAI shouldn’t either, just because their bot can make something that sounds like Pratchett or anyone else. I wrote a short story a while back that my friend said sounded like if Douglas Adams wrote dystopian fiction. Do I owe the Adams’ estate if I were to publish it? The same goes for photography and art. If I take a picture of a pastel wall that happens to have an awkward person standing in front of it, do I owe Wes Anderson compensation? This is how we have to look at it. What’s good for the goose must be good for the gander. I can’t justify punishing AI research and learning for doing the same things that humans already do.