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
I don’t currently have a computer powerful enough to host a top tier LLM like chatgpt4. If I can’t even run it, I sure as shit could never continue to train it with new data. I often use chatgpt with my phone and the thought of doing either one is ridiculous.
There are ways to make money on open source outside of the open source item itself. Redhat has done just that with Linux.
An LLM is just software. No matter what algorithm, tool, or fairy magic was used to amalgamate the data it consumed, they all sucked in open source code and just like any other software that includes open source software, it should be subject to the licensing on the open source software, which pretty much means they should be open source themselves. Companies that want to make money off of AI trained on public data can make their money on the value they add, just like redhat.
The biggest issue I see right now is how to deal with AIs tendency to output data untransformed. Trademark and all those types of licenses are negated as long as the idea within is transformed, but it is really hard to argue transformation when the stupid thing is pooping out word for word quotes, but acting as if it is “new” and transformed.