Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

US Copyright Office found AI companies breach copyright. Its boss was fired

theregister.com

Post reply on HN
Print view View original

111–120 of 410 posts

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#111

Post by jfengel »

Workaccount2 wrote: post #105

I have yet to see someone explain in detail how transformer model training works (showing they understand the technical nitty gritty and the overall architecture of transformers) and also layout a case for why it is clearly a violation of copyright. You can find lots of people talking about training, and you can find lots (way more) of people talking about AI training being a violation of copyright, but you can't fin…

I'm not sure I understand your question. It's reasonably clear that transformers get caught reproducing material that they have no right to. The kind of thing that would potentially result in a lawsuit if you did it by hand.

It's less clear whether taking vast amounts of copyrighted material and using it to generate other things rises to the level of copyright violation or not. It's the kind of thing that people would have prevented if it had occurred to them, by writing terms of use that explicitly forbid it. (Which probably means that the Web becomes a much smaller place.)

Your comment seems to suggest that writers and artists have absolutely no conceivable stake in products derived from their work, and that it's purely a misunderstanding on their part. But I'm both a computer scientist and an artist and I don't see how you could reach that conclusion. If my work is not relevant then leave it out.

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#112

Post by int_19h »

SketchySeaBeast wrote: post #49
Zambyte wrote: post #40

Earlier quoted context omitted.

It took me a while to be convinced that copyright is strictly a bad idea, but these two articles were very convincing to me. https://drewdevault.com/2020/08/24/Alice-in-Wonderland.html https://drewdevault.com/2021/12/23/Sustainable-creativity-po...

The first article is saying that "Copyright is bad because of corporations", and I can kind of get behind that, especially the very long term copyrights that have lost the intent, but the second article says that artists will be happier without copyright if we just solve capitalism first. I don't know about you, but that reads to me like "If you wish to make an apple pie from scratch you must first invent the univers…

The problem of "how do artists earn enough money to eat?" is legitimate, but I don't think it's a good idea to solve it by making things that inherently don't work like real property to work like it, just so that we can shove them into the same framework. And this is exactly what copyright does - it takes information, which can be copied essentially for free by its very fundamental nature, and tries to make it scarce through legal means solely so that it can be sold as if it were a real good.

There are two reasons why it's a problem. The first reason is that any such abstraction is leaky, and those leaks are ripe for abuse. For example, in case of copyright on information, we made it behave like physical property for the consumers, but not for the producers (who still only need to expend resources to create a single work from scratch, and then duplicate it for free while still selling each copy for $$$). This means that selling information is much more lucrative than selling physical things, which is a big reason why our economy is so distorted towards the former now - just look at what the most profitable corporations on the market do.

The second reason is that it artificially entrenches capitalism by enmeshing large parts of the economy into those mechanics, even if they aren't naturally a good fit. This then gets used as an argument to prop up the whole arrangement - "we can't change this, it would break too much!".

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#114

Post by throwaway1854 »

renewiltord wrote: post #79

I wonder when general internet sentiment moved from pro-piracy to IP maximalism. Fascinating shift.

Apples and oranges - and also I don't know if anyone is really supporting IP maximalism.

IP maximalism is requiring DRM tech in every computer and media-capable device that won't play anything without checking into a central server and also making it illegal to reverse or break that DRM. IP maximalism is extending the current bonkers time interval of copyright (over 100 years) to forever. If AI concerns manage to get this down to a reasonable, modern timeframe it'll be awesome.

Record companies in the 90s tied the noose around their own necks, which is just as well because they're very useless now except for supporting geriatric bands. They should have started selling mp3s for 99 cents in 1997 and maybe they would have made a couple of dollars before their slide into irrelevance.

The specific thing people don't want, which a few weirdos keep pushing, is AI-generated stuff passed off as new creative material. It's fine for fun and games, but no one wants a streaming service of AI-generated music, even if you can't tell it's AI generated. And the minute you think you have that cracked - that an AI can create music/art as good as a human and that humans can't tell, the humans will start making bad music/art in rebellion, and it'll be the cool new thing, and the armies of 10Kw GPUs will be wasting their energy on stuff an 1Mhz 8-bit machine could do in the 80s.

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#115

Post by dmoy »

Workaccount2 wrote: post #105

I have yet to see someone explain in detail how transformer model training works (showing they understand the technical nitty gritty and the overall architecture of transformers) and also layout a case for why it is clearly a violation of copyright. You can find lots of people talking about training, and you can find lots (way more) of people talking about AI training being a violation of copyright, but you can't fin…

Not a ton of expert programmer + copyright lawyers, but I bet they're out there

You can probably find a good number of expert programmer + patent lawyers. And presumably some of those osmose enough copyright knowledge from their coworkers to give a knowledgeable answer.

At the end of the day though, the intersection of both doesn't matter. The lawyers win, so what really matters is who has the pulse on how the Fed Circuit will rule on this

Also in this specific case from the article, it's irrelevant?

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#116

Post by dfxm12 »

int_19h wrote: post #44
dfxm12 wrote: post #39

Earlier quoted context omitted.

Please argue in good faith. A new research paper is obviously materially different from "rearranging that text to create a marginally new text".

"Rearranging text" is not what modern LLMs do though, unless you specifically ask them to.

I didn't make this claim. Feel free to bring a cogent argument to a commenter who did.

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#117

Post by biophysboy »

scraptor wrote: post #24
bgwalter wrote: post #13

Earlier quoted context omitted.

The required reasoning is not very deep though: If an AI reads 100 scientific papers and churns out a new one, it is plagiarism. If a savant has perfect recall, remembers text perfectly and rearranges that text to create a marginally new text, he'd be sued for breach of copyright. Only large corporations get away with it.

Plagiarism is not an issue of copyright law, it's an entirely separate system of rules maintained by academia. The US Copyright Office has no business having opinions about it. If a AI^W human reads 100 papers and then churns out a new one this is usually called research.

Having actually done research and published scientific papers, the key limitation is experimentation. Review papers are useful, and AI is useful, but creating new knowledge is more useful. I haven't had much luck using LLMs to extrapolate well beyond their knowledge domain.

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#119

Post by int_19h »

mitthrowaway2 wrote: post #59
evanjrowley wrote: post #52

If AI companies in the US are penalized for this, then the effect on copyright holders will only be slowed until foriegn AI companies overtake them. In such cases the legal recourse will be much slower and significantly limited.

Access to copyrighted materials might make for slightly better-trained models the way that access to more powerful GPUs does. But I don't think it will accelerate foundational advances in the underlying technology. If anything, maybe having to compete under tight constraints means AI companies will have to innovate more, rather than merely push scale.

The problem is that regardless of any innovations, scale still matters. If you figure out the technique to, say, make a model that is significantly better given N parameters - where N is just large enough to be the perfect fit for the amount of training data that you have access to - then someone else with access to more data will use the same technique to make a model with >N parameters, and it will be better than yours.

Re: US Copyright Office found AI companies breach copyright. Its boss was fired

#120

Post by nickpsecurity »

Workaccount2 wrote: post #105

I have yet to see someone explain in detail how transformer model training works (showing they understand the technical nitty gritty and the overall architecture of transformers) and also layout a case for why it is clearly a violation of copyright. You can find lots of people talking about training, and you can find lots (way more) of people talking about AI training being a violation of copyright, but you can't fin…

I did here with proofs of infingement:

https://gethisword.com/tech/exploringai/

Post reply on HN