Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Think of your favorite musicians. How many of them give attribution to where each musical idea came from?

The concept of art as exclusive property is very new. Throughout history, artists have built on one another’s works with no attribution or provenance. It’s really just the past 100 years — Disney, specifically - that have created the cultural mindset that the first person to express something owns it forever and everyone else has to pay them for the privilege of building that next work.

BTW I’m old enough to remember people decrying the rise of desktop publishing (“WYSIWYG”) as the automation of creativity.

I share your disdain for the geriatric political class, but I strongly disagree that this is a situation that needs to be managed. I say we let the arts return to the free for all they were for the fist 80,000 years or whatever.



“ Think of your favorite musicians. How many of them give attribution to where each musical idea came from?”

Certainly not for every individual idea, but good musicians do a lot of attribution. I got to know a lot of music I love now after following a mention on the liner notes of another musician’s album, or having them mentioned in an interview.


Aren’t liner notes the moral equivalent of OpenAI mentioning some source material used for training?

People seem to be asking for much more direct attribution: the pixels in this image are 0.02% from artist X, and 0.006% from artist Y, etc.

It is very rare for a song to include a breakdown of all of the influences that the artist is exercising in that particular piece.


How you are describing that percentage breakdown is how I see this all playing out legally, such that royalty for IP holder = (tags in prompt)/(count of same tags in training data). I am oversimplifying this obviously but you get the idea. This approach would require collective effort of major IP holders but if record labels and streamers can figure out revenue pooling I don't see why it can't work elsewhere.


If the source material was mentioned for every generated image then I think it would be more like what you say. No percentages needed since that's not something we used to get from liner notes either.


But each generated image likely pulls from thousands, maybe millions of pieces of training data, each at a very small weight.


> Think of your favorite musicians. How many of them give attribution to where each musical idea came from?

Great many, if you care to read a bit more of the biographies, autographies, history of music books, interviews, blogs, etc.


At some point we'll probably have insight into learning data of AI. For now copyright makes that super hard.


In what sense does copyright make attribution of that data so hard?

Is it because people are violating copyrights to train these AIs?


They can't publish their training databases because that would be publishing of copyrighted material which is illegal. They can only train which is potential fair use.


It would be more accurate to say that they don't publish their training databases (including sanitized pointers to the copyrighted stuff) because they aren't sure that training is fair use.

They are sure, however, that it is a kind of infringement. Citing "fair use" is an admission of infringement - just a specific kind of infringement that is allowed.


Publishing is definitely not fair use. What's allowed in not an infringement of any kind. Using a right you have cannot be any kind of infringement.


How can training violate copyright? Is reading a book also violation? My understanding was that copyright was about reproduction, not consumption.


It was about unfairly compensated usage, not limited to reproduction.


That doesn’t sound like copyright.


fair enough. It might be better to use other word.


I especially like how Gorillaz artist admits the main hook of their breakaway success song was a rock preset on some niche electronic synthesizer.


I'd be very skeptical that AI would worsen the situation with music. For example, many pop music titles in last decades incorporate the same millennial whoop over and over and over again. I seriously stopped following pop music a long time ago and I can't imagine that AI can make it any more generic if it tried. I don't see a threat for non-generic indie music. AI is good at the generic stuff, as it usually statistically averages out the inputs.


when nirvana played MTV unplugged they mostly played covers from bands that influenced them

also no, disney did not invent the notion of authorship nor royalties. having enough honor not to take credit for someone else's work goes back millennia. attribution goes back millennia, otherwise we wouldn't know the names Sophocles, Aeschylus, Euripides.

Don't get me started on the pharaohs, mother fuckers loved carving their names into things.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: