Destruction of books is normal and totally OK. Publishing houses do it daily in the thousands. When you own something, you can choose how to dispose of it. Come back when you have the title of actually valuable rare books that were bought and destroyed. The 404 article was not well researched in this regard.
The emotional reaction we're conditioned to have hearkens back to 1940s Germany, where the goal was to have people burn all the copies of existing works. But this is not that. They are going for "one of each book" so they have a comprehensive corpus of works to train from. They are willing to buy books to obtain this corpus. It sounds like the system working exactly as intended, with even the book sellers selling these books saying they are "dead inventory".
There's a time to be breathless about "destroying all of humanity's knowledge", but let's wait until it's actually happening.
Again, I'm am very anxious for anyone to come forward with credible information about any valuable/rare/irreplaceable book that was destroyed in this process. So far zero people have been able to do this, probably because they are buying the books in bulk and want to obtain them as cheaply as possible. Book sellers will know the value of any actually rare/valuable books, and would charge accordingly, which would be pointless for e.g. Amazon to pay. They don't need some crazy first edition, they just need some copy to digitize. The incentives just don't add up to "Amazon is buying priceless art and burning it!" the way the narrative suggests.
I really agree with this critique! It's very hard (without hiding behind ToS) to claim that creating the models was A-OK, but distilling them is some sort of ethical breach or attack.
It's interesting to me that the actual motivations for releasing code open-source were very different than I had assumed for the past ~30 years. I always liked to give stuff away so folks could play with and try stuff. As far as I can tell, the only reason not to still give away code is the desire for credit, which was never all that interesting to me.
But then again, I haven't devoted my life to developing and releasing a huge open source project used by millions, so perhaps my opinion should be discounted accordingly.
No, I don't think your opinion should be discounted. It just turns out that a bulk of open-source movement had ulterior motives, and were actually in this for credit or downstream opportunities, not to actually give things away in a pay-it-forward fashion.
The reaction to LLMs pretty clearly revealed who was giving things away, and who was indirectly selling them and using "giving away" for unfair advantage.
> They are destroying the foundations of civilization
No, they are not. For all the hype around that 404 media article, no one came forward with titles or authors. Turns out the booksellers that sold them those books said they were "dead inventory".
If you're going to assert that companies are destroying the foundations of civilization, you're going to have to produce quite a bit more evidence than the (poorly cited) 404 Media article.
Evidence ? Good lord, how about the US district court of justice ? Anthropic physically destroyed millions of books after scanning them - are you really, truly questioning that ?
> Anthropic spent many millions of dollars to purchase MILLIONS of print books, often in used condition. Then, its service providers stripped the books from their bindings, cut their pages to size, and scanned the books into digital form — discarding the paper originals.
Dismissing the above as "dead inventory" is atrocious reasoning. The more obscure/low-circulation books are arguably MORE valuable as training material because they contain information that is less likely to be duplicated across the ordinary web.
And 404 Media's reporting wasn't merely based on a bookseller's terminology of your "dead inventory". There are now unsealed Anthropic internal documents describing Project Panama as an effort to "destructively scan all the books in the world"
They are destroying books that they bought. It should not be controversial that part of ownership is decided how to dispose of the owned object.
But nothing you're citing adds up to "destroying the foundations of civilization". What we have instead is companies that are now paying to get copies of creative works (isn't this what everyone is complaining about with all the "theft" rhetoric?) so they can use them for training.
When they discuss scanning all the books in the world, they are not talking about every copy, since that makes no sense. They want one of each.
So, in the end, this all sounds like some breathless emotional take because large companies are buying copies of books and using them to train AI and throwing them away. They are probably destroying fewer books than the publishing houses do over the same time period.
This is like the water discussion: people with no understanding of agriculture (or publishing) hear a sound-bite about what AI is doing, and, with little-to-no context, devolve into a panic about it.
Again: cite real harm. What books aren't available? What knowledge was destroyed? There are real concerns about AI, but this whole discussion is a distraction.
Many were out of print or were used copies, so they weren't books you could simply order new from a publisher. They are out of print. Anthropic deliberately sourced large quantities through used-book sellers such as Better World Books and World of Books and destroyed the physical copies after scanning them.
If you find this a "breathless emotional take", then I don't even know what to say. Nothing will meet your "evidence" criteria to dislodge your fervent fawning and support for destruction carried out by AI corporations.
> They are destroying books that they bought. It should not be controversial that part of ownership is decided how to dispose of the owned object.
By your reasoning, if I could afford to buy every surviving painting by Van Gogh and Rembrandt and I burned them in a bon fire I wouldn't be taking anything from human culture and nobody should care at all about what I do with my belongings.
Destroying something that is rare or one of a kind and valuable does matter. Doing it so that you can exclusively profit off of charging people for software that regurgitates lossy and error filled versions of what was lost makes it even worse.
In your scenario, the bug is letting someone control all of that in a private collection. But that scenario is completely divorced from what is actually happening, so I don't think it's worth discussing too deeply here.
If you actually object to the concept that people that own a physical object get to decide how to dispose of it, that would be an interesting discussion to have because it would fly in the face of property ownership rights.
Why such a blinkered argument, I own the soda can I just drank but it is still, arguably, wrong of me to chuck it in the trash instead of the recycle bin. Yes, you have the "right" to dispose of things you own but there's more to what's right than that. There's human decency, consideration, awareness of the environment, the spirit of the laws, cases the laws were not designed for in mind etc.
Like, tell your mother or HS teacher you decided not to recycle because you own your trash. See how well that clever argument goes.
you should indeed be able to, but maybe buying art shouldn't be allowed at all, with people only being allowed digital recreations.
why do you need more than a picture of the rembrandt? there's so much good artwork easily available that the rembrandt is only worth money because hoity toity people want to store their money in it and in the future sell it again
> There are now unsealed Anthropic internal documents describing Project Panama as an effort to "destructively scan all the books in the world"
If I do something destructive to my copy of Accelerate -- feed it through a scanner, leave it where the munchkin can reach, whatever -- that does nothing to deprive you of your copy. Books are funny like that; there tend to be lots of identical ones.
Eh, when I read smaller with an integer multiplier, I mentally switch to the reciprocal. Easier than convincing the world not to use "9x smaller". Do you feel the same way about "9x faster"? What you're actually measuring is time, and "faster" is the reciprocal of time, similarly to "smaller" being the reciprocal of size.
yes, actually I do think the phrase 'N times faster' is sensible and logical
if I say 'this Apple M5 chip is 3x faster' than this intel chip, it implies two things:
- the run time of most operations that runs on it is now reduced (so one quantity is smaller)
- but also: MORE WORK is being completed per unit of time compared to the intel chip (so this quantity is greater)
So yes, a greater quantity is being measured in the apple chip compared to the intel when you say apple is N times faster.
i guess this a quirk with the word 'faster' - it actually measures two things, time and work performed per unit of time. the word smaller just measures size.
like, if I give a customer a cup of coffee one day, then give them a SMALLER cup of coffee the next day, for the same price, but declare "It's now 2X more space efficient!" i.e. it's now half the size, I'm certain the customer is gonna be pissed.
I feel like this becomes less clear if you're talking about "one-time" events. "Our website loads 2x faster now" doesn't mean that they are sending more data per second, and the user won't be doing more website loads. The website load time going from 1 second to 0.5 seconds is "2x faster" and I think that that is a very common way of communicating such performance increases too.
Ah, but in this case the customer is buying the coffee for the caffeine content. And you've doubled the caffeine content per volume! Twice as efficient a delivery mechanism, similar to this model!
I've had the same uncomfortable revelation. I probably read Sloshdot for close to 20 years before dropping it exclusively for HN, but now I have it back in my RSS feed and it is consistently producing interesting articles that I would otherwise miss.
Anti-intellectual? Because they don't believe claims that are backed up only by untested hypotheticals?
If we're going to have whistleblowers emerge declaring a 10% chance we all die because of AI, we're going to need more than vibes. Real experts can explain their position plainly. That has not happened here.
I read Bostrom's Superintelligence: Paths, Dangers, Strategies in 2014. It was interesting, but not terribly persuasive. Does If Anyone Builds It, Everyone Dies contain arguments that would be new to me after reading Superintelligence?
I would personally expect so, but your mileage may vary. Controlling for the fact that you didn't find the former persuasive, my first thought would be that the latter might help. Among other things, it is much more specific and detailed in various ways, not least of which being informed by an additional 11 years of history. The former was not written in a world of pervasive AI, or with current AI architecture in mind; the latter has that context.
You said "Real experts can explain their position plainly.". The experts have explained their position plainly. If you disagree, you should be able to explain your position plainly; "feels like vibes to me" is not an explanation, it's a feeling.
If you want to argue that there's something invalidating the arguments, by all means do. But it's really tiring to get many variations of "nobody's made the case for it!" or "there's no evidence for it!" or "vibes!".
My position is that there are too many off-ramps between us and the proposed inevitability. And the argument is not symmetric. The null hypothesis is that the world continues to function. The burden of proof is on the authors of the book to show that human destruction is an inevitability as the result of this technology, and they have failed to show that.
This is cited in multiple reviews of the book in reputable publications. The book simply fails to compel belief.
> Not a single one of you self-proclaimed experts has actually debunked the technical evidence I provided because you cannot.
This isn't a great way to start a conversation.