> aren’t copyrighted so this is actually more ethical that’s what OpenAI did
why is that? all works, as long as it is "original" should be granted copyright. It should belong to the person who made it (in this case, the user, not openAI).
why is that? all works, as long as it is "original" should be granted copyright. It should belong to the person who made it (in this case, the user, not openAI).
If I say to one of my human friends "write me a nursery rhyme", the copyright of the resulting rhyme would obviously belong to my friend - despite me prompting them. Clearly the prompt itself does not universally count as "making" it.
Let's say I made a "NovelSnippetAI", which contains a corpus of prewritten material. You can prompt it, and it will return a page which matches the sentiment of your prompt best. I think we can agree that the copyright of the page will still belong to the original writer - the user only did a query.
What if I did "NovelMixAI", which did exactly the same but alternated lines from the two best matches? What about "NovelTransformAI", which applied a mathematical formula to the best match and fed the output to a fixed Markov Chain? Now we're suddenly at "LLMAI", which does the same using a neural network - what makes it different from the rest?
Long-standing precedent is that any automated work does not qualify for copyright. You can only copyright human work.
This is the wrong analogy. The LLM is more like Photoshop and the prompt little more than a filter configuration. A machine cannot copyright its own output but a human guiding that machine can.
thats the contention, its not a person who's made it.