Those claiming AI training on copyrighted works is "theft" misunderstand key aspects of copyright law and AI technology. Copyright protects specific expressions of ideas, not the ideas themselves. When AI systems ingest copyrighted works, they're extracting general patterns and concepts - the "Bob Dylan-ness" or "Hemingway-ness" - not copying specific text or images.
This process is akin to how humans learn by reading widely and absorbing styles and techniques, rather than memorizing and reproducing exact passages. The AI discards the original text, keeping only abstract representations in "vector space". When generating new content, the AI isn't recreating copyrighted works, but producing new expressions inspired by the concepts it's learned.
This is fundamentally different from copying a book or song. It's more like the long-standing artistic tradition of being influenced by others' work. The law has always recognized that ideas themselves can't be owned - only particular expressions of them.
Moreover, there's precedent for this kind of use being considered "transformative" and thus fair use. The Google Books project, which scanned millions of books to create a searchable index, was ruled legal despite protests from authors and publishers. AI training is arguably even more transformative.
While it's understandable that creators feel uneasy about this new technology, labeling it "theft" is both legally and technically inaccurate. We may need new ways to support and compensate creators in the AI age, but that doesn't make the current use of copyrighted works for AI training illegal or unethical.
For those interested, this argument is nicely laid out by Damien Riehl in FLOSS Weekly episode 744. https://twit.tv/shows/floss-weekly/episodes/744
No, but they're saving money which is effectively the same thing. There's no practical difference between earning 50 bucks and getting a 50 buck discount.
That’s not quite true, though, is it?
$50 earned is yours to spend on anything. A $50 discount is offered by a vendor to entice you to spend enough of your money on them to make the discount worthwhile.
Pirates don’t pirate because they’re trying to save money on something they would have bought otherwise… typically they pirate because the amount they consume would bankrupt them if they purchased it through legitimate means, so they would never have been a paying customer in the first place.
So, if they wouldn’t have bought it anyway, and they’re not reselling it, did they really harm the vendor? Whether they pirated it or not, it wouldn’t affect the vendor either way.
That’s not really the same thing, in my opinion.
If you were able to pay for everything handily but pirated anyway, or if you resold pirated content, then yeah you have something similar to theft going on. But that’s not really the norm; those people are doing something bad irrespective of the piracy itself, aren’t they?
They wouldn't have bought all the content they have pirated if piracy was not an option but they would have bought some of it.
Piracy has saves money. Saving money means I have more money to spend on other things. Earning money means I have more money to spend on other things. There's no practical difference between the two.
In my view, my point still stands; being against one but not the other is hypocritical.
It’s not hypocritical if you believe that theft is wrong because it hurts another person, rather than wrong because you don’t deserve the thing or that it offers you an unfair advantage. Your argument leans heavily on the latter but mine the former.
Then please explain how pirating movies or games doesn't cause harm to other people but training AI with the copyrighted work of others does.
In both cases you're taking something for free that you're expected to pay for. In both cases there's someone not getting paid. The only difference between the two that I can see is the scale which is irrelevant from the point of view of the argument I'm making which is that it's hypocritical to be against one but not the other.