[jstummbillig]

Forever preserving the content of a rare book by consuming one (1) copy is the kind of thing we should always be doing more of.

Roughly nobody would have had access to that copy. Roughly everyone can now benefit from its content.

[28304283409234]

One (1) company destroys and (illegally?) copies a rare book. This knowledge is now of that company, not of humanity. Other companies will feel the need to do the same and before you know it knowledge is walled off in the gardens of the AI companies, and the rare books are gone and destroyed.

[jstummbillig]

When are you or anyone more likely to get access to a rare book:

a) When a company has made a digitized copy of it

b) When a few previous hard copies are kept somewhere in the world

Idk what Anthropic plans to do with this, but note that my original comment was not about Anthropic at all. I simply think it's a good tradeoff to destroy a few copies of rare books if that helps digitizing them. I am not against simply legally forcing companies that do this to release those copies in due time. In the meantime I am also happy to have them as part of the LLM corpus.

> illegally?

No, copying rare books is very likely not illegal, because, presumably, they are quite old.

[pfdietz]

The scan doesn't go away, and can be released when the copyright expires.

Oh no, a lump of cellulose is gone. Will no one think of the fibers?

[collinmcnulty]

> can be released when the copyright expires

But it ... won't be? The companies have no motivation to do so. The copyright on lots of these books are surely already expired, so if they wanted to they could be putting these up now. I'm sure internally this is viewed as a corpus of knowledge they have that their competitors do not, so they will not release them unless something forces them to.

[pfdietz]

Yes, without any justification let's assume the worst. This is true good faith arguing.

[polarbearballs]

I have a deep rooted doubt that "roughly everyone can now benefit from its content" If there's any value in the content it will immediately be monetized with an aggressive pay structure and had it sat in a library, that same value would be free.

[anon373839]

> Roughly everyone can now benefit from its content.

You mean, roughly everyone who uses that particular model provider’s products. For truly rare books, this has an anticompetitive flavor, since it ensures others can’t train models from the same knowledge.

[maccard]

Where can I read any of these destroyed copies from anthropic?

[XorNot]

What do you imagine a rare book is? It's not the only extant copy of the writing in that book, in fact it's not even like it represents a significantly limited accessibility of that book.

[breezybottom]

It's going to be hard for you to do more of that, considering that AI forms are destroying the books. You know, the whole point of the article...

[Certhas]

Roughly Anthropic owners benefit from the content, unless the scans are made available somehow.

[Sha1rholder]

> Roughly everyone can now benefit from its content.

You mean every Anthropic?