4 ms·
Creators Suing AI Companies over Copyright Further Entrenches Big Tech
- glimshe 3y agoThe chances of creators beating Big Tech on this one? Zero. Just see how the incentives are aligned. If the creators win, Big Tech has to pay millions/billions to them and reduce their profit margins. If creators lose, Big Tech can provide even more money to political donations. This said, even if this wasn't the case, I don't think creators have the moral/legal standing to win even if the decision was entirely fair given many arguments discussed on HN and elsewhere.
- buildbot 3y agoBig tech would _love_ to "loose" these cases so they can gain a complete monopoly on it. It would take massive capital to pay the aggregate license holders, working in Capital's favor.
- glimshe 3y agoNot paying will be even cheaper. They don't need this to obtain the de facto monopoly given the massive costs in training. Even if other companies offer their own AI, it will be for niche scenarios (like character creation for games), and training will use Big Tech's cloud resources. So they will corner general purpose AI and will extract hefty fees for special cases. If creators win, many companies will be unable to cover niches (which are unprofitable for Big Tech but not smaller companies, and will become more so with creator fees) and thus fewer resources will go to cloud services.
- buildbot 3y ago> Not paying will be even cheaper. They don't need this to obtain the de facto monopoly given the massive costs in training. Even if other companies offer their own AI, it will be for niche scenarios (like character creation for games), and training will use Big Tech's cloud resources. The massive costs of training is why they will pay, in order to lock out others. If you spend _billions_ in training, who cares about a 10 million dollar dataset? The rest of you argument reads more like a push for exactly what the article calls for. Many companies will not be able to amass large datasets to attempt to compete with large companies if they have to pay for it. What happens if Cool AI Startup spends 50% of their funding on a dataset, but it's useless after months of experiments? Large companies can easily absorb those costs, small companies can't.
- gumballindie 3y ago[flagged]
- buildbot 3y agoYep exactly - large corporations are like, 25-75% compliance keeping machines before they do anything else. They have entire departments for this kind of thing. All these companies would love to lock the best training data behind a multi-million dollar licensing fee. Let information be free. Don't try to create artificial scarcity on digital goods though. Including art. If you want to sell your art, you are more than welcome too! But expecting a company to send you a payment because they read your publicly posted fanfiction seems likely to fail.
- mvdtnz 3y agoThere's a lot of assumptions and assertions in this article and very few sources or links. Not even sure where to start with this one because almost everything here is just an assertion from the author. > However, because so many people think that they’re supporting creators and “sticking it” to Big Tech in supporting these copyright lawsuits over AI, I don't care about "sticking it to big tech", I think it's only fair to pay for using someone else's work to your own benefit. > Copyright law, by its very nature as a monopoly right, has always served the interests of gatekeepers over artists. Copyright has never served artists? That's a huge assertion to just throw out there. > This is why the most aggressive enforcers of copyright are the very middlemen with long histories of screwing over the actual creatives: the record labels, the TV and movie studios, the book publishers, etc. These are the groups enforcing copyright because in a lot of cases they are the copyright holders. > This is why we’ve said that AI training will never fit neatly into a licensing regime. Certainly true - we will need a new licensing regime to handle this. It's a brave new world. > And, because of the nature of training data, which only needs to be collected once, it’s not likely that this will be a recurring payment, but a minuscule one-off for the right to train on the data. This is a very strange assertion. Spotify only needs to download your band's MP3 files once so they should only pay you once? No, they use it on an ongoing basis to serve their customers, so recurring payments are appropriate. > But, given the enormity of the amount of content, and the structure of this kind of thing, the cost will be extremely high for the AI companies (a few pennies for every creator online can add up in aggregate), meaning that only the biggest of big tech will be able to afford it. Yeah this will lock out all of those scrappy mum and pop shops training their multi-billion parameter models on the entire internet. I don't claim to know the proper solution to this problem, but it does feel desperately unfair to me to see companies like OpenAI slurping up the output of artists and benefitting greatly from it while the artists get nothing in return. I don't know if copyright law is the right place to look or not. Like I said, it's a brave new world. But this article is not convincing to me, at all.
- kianlocke 3y ago> I think it's only fair to pay for using someone else's work to your own benefit. How many artists do you know that pay out a portion of the proceeds of any of their works to the artists that have influenced and inspired them? Why should we have a separate set of laws for men and machines? Why would we assume that any algorithms used to determine 'inspirational ancestry' will only ever be used against machines? This is a power grab, pure and simple, an attempt to add style to the list of copyrightable immaterials -- Disney 2.0, but on a much more massive and debilitating scale.
- sgt101 3y agoBecause they are good at stealing doesn't mean that people shouldn't try and stop them stealing or that it will be a good outcome for anyone if they successfully steal all the stuff.
- rvz 3y agoSurely this means it is perfectly fine to train on people's voices without their permission and then claim the generated songs (with their voices) as your own and use the 'fair use' excuse just because paying for a license means it will make the grift too expensive for large tech companies and extremely costly for smaller ones? I don't think so. Given that Runway partnered with Getty Images with a licensing deal [0], and the same with Meta partnering with musicians to clone their voices (with permission and compensation) [1] it only tells us that is where it is going and how the Stability AI v Getty lawsuit will settle with such a deal rather than attempting to violate the copyright of the author and then nonsensically claim 'fair use' with their commercial offering. [0] https://www.axios.com/2023/12/04/runway-getty-images-ai-video https://www.axios.com/2023/12/04/runway-getty-images-ai-vide... [1] https://www.zdnet.com/article/meta-just-entered-the-ai-chatbot-race-with-three-major-product-announcements/ https://www.zdnet.com/article/meta-just-entered-the-ai-chatb...
- kianlocke 3y agoJust curious -- do you feel that a person who learns to perfectly imitate (do an impression of) another's voice should be required to pay them for any performances done in that voice, if they do not claim to literally be that person?
- wokwokwok 3y ago> while the actual creators get pennies. … > And then all of the actually interesting smaller companies and open source models are screwed. The argument here is totally disingenuous. You can tldr it like this: Either a) you pay people for their content, which makes things prohibitively expensive for small companies. Or b) you can screw artists over (again) and give them literally nothing so you enable small companies to use and create AI models with legal protections. Now, those are legitimate considerations; you can argue that for society as a whole having lots of small companies is better than having a few big ones. …but you can’t help but look at that outcome and go: What about the artists? Do they get nothing? Is that ok? Is it not a little bit greedy and selfish to say you take everything, pay nothing and screw them over? Spotify and Apple Music aren’t great; they don’t pay well. …but it’s something right? How can any reasonable person sit there and say it would be better to have 20 different Spotify clones that paid no one anything other than their share holders? It’s greedy and selfish. People deserve to be paid for their work.
- pixl97 3y ago>People deserve to be paid for their work. All fun and games until we're long past the point of AI having to learn from people. Then what do you plan on doing? Tax corporations, provide UBI. Thinking that we'll need vast workforces in the future is folly.
- wokwokwok 3y ago> Then what do you plan on doing? I dunno; but what I do know, is that being selfish and greedy is wrong. It's wrong now. It'll be wrong when our AI overlords rule over us all. How about we talk about meaningful, real world things that are happening right now instead of arbitrary nebulous future possibilities? People, doing work, right now, deserve to be paid for that work, because there is no UBI that currently exists.
- pixl97 3y agoThen unionize. If you're demanding as a bunch of individuals that big corporations pay you, what you're going to get is big corporations fighting you against each other for slave wages, and the big corporations owning almost all human creative output in the future.
- comfysocks 3y agoThe moment OpenAI can find a way to get superior results without training on copyrighted materials, you will probably see them being more vocal about the issue as it will give them a moat. This is one of the ways you know they are training on copyrighted materials.