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

Meta pirated books to train its AI

theatlantic.com

Post reply on HN
Print view View original

31–40 of 77 posts

Re: Meta pirated books to train its AI

#31

Post by wwweston »

HPsquared wrote: post #20
wwweston wrote: post #19

This is one of the reasons there should simply be a big tax on AI profits, assuming they come to pass: they literally require uncompensated training on the sum of human intellectual effort. There is no AI without standing on the shoulders of both giants and thousands of everyday authors. If you're just giving the result back to humanity, OK, there's a case for it being a fair trade, though there's still the question…

Tax doesn't reward the authors. Especially foreign authors.

It could be structured to reward living authors, mechanical licensing has worked this way.

But even if it wasn't -- even if couldn't be -- you could use it to fund all kinds of public goods which would benefit living authors. Maybe even a basic income if it turns out to be successful enough.

It's a different pro-social bargain than direct compensation for the fruits of labor, but at least it's still a bidirectional one.

Re: Meta pirated books to train its AI

#32

Post by daniel_reetz »

vintermann wrote: post #21
upghost wrote: post #5

There seems to be a precedent with tech companies that if you do a lot of illegal stuff in the beginning you can count on either failing before it matters or you can buy your way out of the consequences. Bonus points if you induce regulatory capture to make sure no one else can follow in your footsteps. Please HN do your thing and prove me wrong or tell me the benefits to society outweigh the cons, because otherwise…

Well, the fruit of it, Llama 3, is public. Maybe it comes with some sort of license, but considering how it was made, I wouldn't feel guilty about violating it. Then again, I would happily train a model on Anna's Archive myself without feeling guilty about it, if I had the resources to do so. I think it ranks very, very low on the list of bad things Facebook has done.

The similarities to Google Books are interesting. Google created Books based on their own scans of books, which publishers considered piracy.

The legal battle initiated by the Authors Guild was vicious and it damaged and hampered Books and other entities and scanning projects like the Internet Archive.

Now Facebook pirates all the world's books, uses them without paying authors or publishers, and seemingly faces no consequences.

I would strongly prefer to live in a world where we could easily search and access all books, than one where the richest guys get to exploit them without consequences.

Re: Meta pirated books to train its AI

#34

Post by AshamedCaptain »

JTyQZSnP3cQGa8B wrote: post #13
upghost wrote: post #5

There seems to be a precedent with tech companies that if you do a lot of illegal stuff in the beginning you can count on either failing before it matters or you can buy your way out of the consequences. Bonus points if you induce regulatory capture to make sure no one else can follow in your footsteps. Please HN do your thing and prove me wrong or tell me the benefits to society outweigh the cons, because otherwise…

I can't prove you wrong and I don't think there is any benefit to society when a company with a black box doesn't give back to the community (aka humanity as a whole) and doesn't pay anyone. Copyright may be broken, but the only realistic way to encourage creators is to give them money. We don't live in a utopia, and people need money to eat. Also, in France and as an individual, if I were to openly torrent all the c…

> I would have to spend the rest of my life in jail whereas Zuck can enjoy having more billions with the same behavior.

Or you would driven by suicide for much smaller violations - https://en.wikipedia.org/wiki/Aaron_Swartz

I mean, one could even argue that what Aaron was doing was not even violating copyright, as he had access....

Yet this guy is dead, and the Facebook owners are instead going to get a nice raise.

Re: Meta pirated books to train its AI

#35

Post by binarymax »

stackedinserter wrote: post #15

I "pirated" (read) books to train myself. I guess I owe authors a percentage of my income now.

Let’s look at this like food, instead of knowledge. You ate a donut, and you got energy. Maybe you’ve eaten 5 donuts per year, for 20 years…so a hundred donuts. Maybe you didn’t pay for all of them, you got some at work or at a friends house.

They stole a billion donuts.

Re: Meta pirated books to train its AI

#38

Post by simonw »

Here's a really interesting line from that story:

> In a message found in another legal filing, a director of engineering noted another downside to this approach: “The problem is that people don’t realize that if we license one single book, we won’t be able to lean into fair use strategy”

I hadn't heard that idea before. Any IP law experts able to give useful context on that one?

Re: Meta pirated books to train its AI

#39

Post by Kapura »

upghost wrote: post #5

There seems to be a precedent with tech companies that if you do a lot of illegal stuff in the beginning you can count on either failing before it matters or you can buy your way out of the consequences. Bonus points if you induce regulatory capture to make sure no one else can follow in your footsteps. Please HN do your thing and prove me wrong or tell me the benefits to society outweigh the cons, because otherwise…

No, you're just right, and you should feel upset about it. Think about the systems and incentives that create this outcome. Think about what needs to change for better outcomes to be possible. Find people and work towards those goals.

Re: Meta pirated books to train its AI

#40

Post by xyzzy9563 »

dartos wrote: post #14
xyzzy9563 wrote: post #8

Earlier quoted context omitted.

Downloading pirated material isn't illegal in the U.S., sharing it is.

Really dude? I think a solid argument can be made that facebook downloaded and shared pirated material with literally every ML employee they had at the time… EDIT: to those who downvoted, why? Do you disagree with what I said?

Sharing something locally doesn't fall under federal copyright law. They don't have jurisdiction.
Post reply on HN