2 ms·

Opinion: The Copyright Office is making a mistake on AI-generated art

7 points by kristianp 3y ago

necovek 3y ago
> Photographers would have to submit detailed descriptions of how they staged the scenes they photographed. ... > This would be a waste of time for everyone involved. And the same is true for AI-generated art. An article on copyright not even covering the rationale for having copyright in place? Copyright is there to protect investment in creative efforts in the time where making a copy is cheap. If there was enough work and investment in a photo or AI generated image, it makes sense to protect it and go through the registration process. As such, it would not be a waste of anyone's time, but would instead stop non-effort works from being copyrighted because people wouldn't bother — which is exactly how it should be.
hilbert42 3y ago
"I don’t think these more recent decisions are going to age well." Perhaps not, but what it does signal is that copyright and patent law as they now stand are well behind the times and have lost relevancy and urgently need revision. Recent advances in AI have made this glaringly obvious. If, say, an artist, programmer or designer seeds an AI with an idea how much does that seeded idea account towards the final created work whatever type of work that may be—whether it be a painting or other decorative work of art, or software, or a circuit, or circuit board design, or a machine part, etc., etc? Arguments about how significant the seed is, its extent, its relevance and how much ancillary information or supplementary parameters/data are added to the computation become highly relevant. The loghic extends—how far does one take mechanical nonhuman aids back and still consider them relevant to copyright? This argument can even be extended back to what extent the artist's brush—a nonhuman inanimate object—contributes to an artist's work. Should the designer of, say, a special type of patented paintbrush be paid a part of the artist's royalties for without said brush the artist wouldn't have been able to produce the copyrighted work (that brought him/her income). That logic extends even further: take artists' education and training for without this input from others (and society) they wouldn't be able to create their works. In most cases artists have been trained at the expense of the state not to mention others. Restated, how much of an artist's royalties does the artist owe to the state and his/her teachers? What does he owe back to society or what obligation does he have to share some of his/her royalties with others? If, say, under new copyright law that artists can only claim part of the royalties of a work they generate under AI, then such a law must be inconsistent unless other protagonists who also helped artists create works also share in the royalties. Making such a law fair would also mean that the developer of the patented bush, artists' teachers including the state would also be owed partial royalties. In essence, copyright would no longer be an exclusive right under law; to be fair copyright would then have to become a shared right. It's clear to me that under current copyright law that the Copyright Office hasn't made a mistake in this ruling because it understands many of the issues I've raised and is well aware of the problems and limitations of current copyright law. There's little doubt in my mind that both AI and the information age, IT etc., are going to force copyright reform no matter what rights holders think or how much they oppose such reform. Without major reform the whole copyright/patent system will break down as technology develops even further.
xg15 3y ago
I think in some sense, I'd appreciate if AI art would be made copyrightable, because it would force the courts to evaluate AI art with the same framework that is used for other creative processes, exposing the absurdities within: I.e., if I use an AI model trained on existing art and get a result, is my art then a derived work of that existing art? If yes, of which one exactly? If no, can I use AI models to effectively strip copyright from existing works?

7 points by kristianp 3y ago

> Photographers would have to submit detailed descriptions of how they staged the scenes they photographed. ... > This would be a waste of time for everyone involved. And the same is true for AI-generated art. An article on copyright not even covering the rationale for having copyright in place? Copyright is there to protect investment in creative efforts in the time where making a copy is cheap. If there was enough work and investment in a photo or AI generated image, it makes sense to protect it and go through the registration process. As such, it would not be a waste of anyone's time, but would instead stop non-effort works from being copyrighted because people wouldn't bother — which is exactly how it should be.

hilbert42 3y ago

"I don’t think these more recent decisions are going to age well." Perhaps not, but what it does signal is that copyright and patent law as they now stand are well behind the times and have lost relevancy and urgently need revision. Recent advances in AI have made this glaringly obvious. If, say, an artist, programmer or designer seeds an AI with an idea how much does that seeded idea account towards the final created work whatever type of work that may be—whether it be a painting or other decorative work of art, or software, or a circuit, or circuit board design, or a machine part, etc., etc? Arguments about how significant the seed is, its extent, its relevance and how much ancillary information or supplementary parameters/data are added to the computation become highly relevant. The loghic extends—how far does one take mechanical nonhuman aids back and still consider them relevant to copyright? This argument can even be extended back to what extent the artist's brush—a nonhuman inanimate object—contributes to an artist's work. Should the designer of, say, a special type of patented paintbrush be paid a part of the artist's royalties for without said brush the artist wouldn't have been able to produce the copyrighted work (that brought him/her income). That logic extends even further: take artists' education and training for without this input from others (and society) they wouldn't be able to create their works. In most cases artists have been trained at the expense of the state not to mention others. Restated, how much of an artist's royalties does the artist owe to the state and his/her teachers? What does he owe back to society or what obligation does he have to share some of his/her royalties with others? If, say, under new copyright law that artists can only claim part of the royalties of a work they generate under AI, then such a law must be inconsistent unless other protagonists who also helped artists create works also share in the royalties. Making such a law fair would also mean that the developer of the patented bush, artists' teachers including the state would also be owed partial royalties. In essence, copyright would no longer be an exclusive right under law; to be fair copyright would then have to become a shared right. It's clear to me that under current copyright law that the Copyright Office hasn't made a mistake in this ruling because it understands many of the issues I've raised and is well aware of the problems and limitations of current copyright law. There's little doubt in my mind that both AI and the information age, IT etc., are going to force copyright reform no matter what rights holders think or how much they oppose such reform. Without major reform the whole copyright/patent system will break down as technology develops even further.

xg15 3y ago

I think in some sense, I'd appreciate if AI art would be made copyrightable, because it would force the courts to evaluate AI art with the same framework that is used for other creative processes, exposing the absurdities within: I.e., if I use an AI model trained on existing art and get a result, is my art then a derived work of that existing art? If yes, of which one exactly? If no, can I use AI models to effectively strip copyright from existing works?