Remix.run Logo
eaglelamp a day ago

If we are going to have a general discussion about copyright reform at a national level, I'm all for it. If we are going to let billion dollar corporations break the law to make even more money and invent legal fictions after the fact to protect them, I'm completely against it.

Training a model is not equivalent to training a human. Freedom of information for a mountain of graphics cards in a privately owned data center is not the same as freedom of information for flesh and blood human beings.

r3trohack3r a day ago | parent | next [-]

You’re setting court precedent that will apply equally to OpenAI as it does to the llama.cpp and stable diffusion models running on your own graphics card.

photonthug a day ago | parent | next [-]

I don’t know about that, we seem to be so deeply into double standards for this stuff that we’ve forgotten they are double standards. If I aggressively scrape content from anywhere and everywhere ignoring robots.txt and any other terms and conditions, then I’ll probably be punished. Corporate crawlers that are feeding the beast just do this on a massive scale and laugh off all of the complaints, including those from smaller corporations who hire lawyers..

darioush a day ago | parent | next [-]

oh they hate it so much when this hypocrisy is pointed out. better put the high school kids downloading books on pirate bay in jail but I guess if your name starts with Alt and ends in man then there's an alt set of rules for you.

also remember when GPU usage was so bad for the environment when it was used to mine crypto, but I guess now it's okay to build nuclear power plants specifically for gen-ai.

FeepingCreature 17 hours ago | parent | prev [-]

Great, let's legislate corporate liability for excessive data use from crawlers. I'm fully there with you.

munificent a day ago | parent | prev | next [-]

SGTM.

Honestly, seriously. Imagine some weird Thanos showed up, snapped his fingers and every single bit of generative AI software/models/papers/etc. were wiped from the Earth forever.

Would that world be measurably worse in any way in terms of meaningful satisfying lives for people? Yes, you might have to hand draw (poorly) your D&D character.

But if you wanted to read a story, or look at an image, you'd have to actually connect with a human who made that thing. That human would in turn have an audience for people to experience the thing they made.

Was that world so bad?

autoexec 20 hours ago | parent | next [-]

Imagine a world where Thanos snapped his fingers and photoshop (along with every digital application like it) was wiped from Earth forever. The world would keep on turning and artists would keep on creating, but creating art would be more difficult and fewer people would be able to do it (or even touch up their own photos).

Would that world be so bad? Was the world really so horrible before photoshop existed?

What if we lost youtube? What if we lost MP3s?

We could lose a lot of things we didn't always have and we'd still survive, but that doesn't mean that those things aren't worth having or that we shouldn't want them.

FeepingCreature 17 hours ago | parent | prev | next [-]

That world was worse. It wasn't much worse, because we haven't seen most of the benefit of GenAI yet, but yes I would say that it was worse.

It wasn't "so bad", but any history of improvement can be cut into slices that aren't "so bad" to reverse.

dcow a day ago | parent | prev [-]

Obviously the former status quo wasn’t that bad. But the opposite is also true, AI democratizes access to pop culture. So now when I connect with a human it’s not to share memes, it’s higher order. IOW we can spend more time playing D&D because we didn't have to draw our characters.

munificent a day ago | parent [-]

> AI democratizes access to pop culture.

Pop culture was already democratized. That's literally what makes it popular culture.

> So now when I connect with a human it’s not to share memes, it’s higher order.

I suspect that improving the image quality of the memes does not measurably improve the quality of the human connection here.

> IOW we can spend more time playing D&D because we didn't have to draw our characters.

You never had to draw your characters. You can just play and use your imagination. Why would we let LLMs do our dreaming for us?

dcow a day ago | parent [-]

It's a rhetorical example. Suppose you need to create an avatar of your character. Why does it follow that it's not beneficial to have an AI help generate the avatar?

You're responding to the specific example, not the general argument. Unless your counter is that whatever humanity is doing that AI is helping is probably stupid and shouldn't be done anyway.

munificent a day ago | parent | next [-]

> Unless your counter is that whatever humanity is doing that AI is helping is probably stupid and shouldn't be done anyway.

No, my counter is that whatever generative AI is doing is worth doing by humans but not worth doing by machines.

As the joke comic says: We thought technology was going to automate running errands so that we had time to make art, but instead it automates making art while we all have to be gig workers running errands.

nradov a day ago | parent | prev [-]

No one needs an avatar. You can draw a stick figure or take a selfie or whatever. This is all so silly and trivial.

dcow a day ago | parent [-]

Consider consulting documentation then. A model can help sift through orders of magnitude more literature than you can in the same timeframe.

nradov a day ago | parent [-]

OK? What does that have to do with pop culture IP rights?

If you're building an LLM for management or technical consulting then the valuable content is locked up behind corporate firewalls anyway so you're going to have to pay to use it. In that field most of what you could find with a web crawler or in digital books is already outdated and effectively worthless.

codedokode a day ago | parent | prev [-]

Can stable diffusion be created without using copyrighted content? Maybe we should have some exemption for non-commercial research but definitely not for commercial exploitation or generating copyrighted images using open-source models.

godelski 21 hours ago | parent | next [-]

There is already exemptions for research. Look at licensing around things like ImageNet. There's similar licensing around things like LAION and Common Crawl[0] It's also not legal to just scrape everything without paying. There's a reason the NYT sued OpenAI and then got a settlement. It's still illegal for Meta to torrent terabytes of textbooks too.

[0] https://commoncrawl.org/terms-of-use

  > In this regard, you acknowledge that you may not rely on any Crawled Content created or accumulated by CC.  CC strongly recommends that you obtain the advice of legal counsel before making any use, including commercial use, of the Service and/or the Crawled Content.  BY USING THE CRAWLED CONTENT, YOU AGREE TO RESPECT THE COPYRIGHTS AND OTHER APPLICABLE RIGHTS OF THIRD PARTIES IN AND TO THE MATERIAL CONTAINED THEREIN.
FeepingCreature 17 hours ago | parent | prev | next [-]

Can an artist be created without using copyrighted content? Raise a child without movies, books, songs or the internet, see how much they contribute to "popular culture".

codedokode 14 hours ago | parent | next [-]

There is liberally licensed content like creative commons.

FeepingCreature 14 hours ago | parent [-]

I love CC but culturally it's a nonfactor.

sksrbWgbfK 15 hours ago | parent | prev [-]

Neglect is illegal and I don't understand your point.

CaptainFever 21 hours ago | parent | prev [-]

Public Diffusion: https://source.plus/public-diffusion-private-beta

robocat a day ago | parent | prev [-]

> invent legal fictions after the fact

You're reading into the situation...

For the US getting legislators to do anything is impossible: even the powerful fail.

When a legal system is totally roadblocked, what other choice is there? The reason all startups ask forgiveness is that permission is not available.

(edit). Shit. I guess that could be a political statement. Sorry