From the website of the company that’s destroying books to feed them into the ravenous AI maw.

[msbellows@c.im]

@overholt Also from the article: "Anthropic described the project as 'our effort to disruptively scan all the world’s books' in an internal memo, which noted 'we do not want it to be known that we are pursuing this project'."

[drwho@masto.hackers.town]

@ashwin @datarama @angiebaby @overholt "Fungible capital human assets," at a few places.

[vfrmedia@social.tchncs.de]

@aehoogestraat @overholt goose is valued by cooking it (and then wondering why there isn't any more supply of golden eggs)

[thvv@social.vivaldi.net]

@overholt
this is like saying that a human
being is an arrangement of
oxygen, hydrogen, carbon, and
other elements.

[angiebaby@mas.to]

@whoosh @overholt

Missed the point entirely, but I tend to expect that from mansplainers.

[riley@toot.cat]

@Fragarach There's precedent: Mireille Breitwieser.

Strangely, despite the initial assessments, it seems like the 2025 Louvre heist's gang didn't rush to melt their loot down into nameless gold.

@overholt

[riley@toot.cat]

@Guillotine_Jones It's also probably genAI-generated, reaching a previously unreachable level of 'self-serving'.

@overholt

[msbellows@c.im]

@riley @Guillotine_Jones @overholt Actually, that language pretty much comes from dicta in the federal court decision allowing one AI company to settle copyright infringement claims for $1.6 billion. That decision said that the only way around such claims is to destroy the physical book after extracting the information, because then you're not stealing, you're transforming. An absolutely unintended consequence on the court's part, but the AI companies all are running with it.

[gladsdotter@mastodon.world]

@overholt I am hanging on to all of my print reference books so that when AI garbles the content of its ingested books, I still have a touchstone for reality.

[raymierussell@mastodon.scot]

@thvv @overholt

It sounds like what late Philosopher Daniel Dennett would have called a "deepity".

from wikipedia

"(a deepity is) a statement that has two meanings: one that is true but trivial, and another that sounds profound and would be important if true, but is actually false or meaningless. "

[endorama@hachyderm.io]

@overholt this is so horrible, what a fucked up timeline. 🫂

[bluegreenandfree@mastodon.energy]

@overholt This is just disgusting. Really underlines that LLMs are illiteracy machines.

These tech companies are half a step away from just being Futurama's Brain Spawn with their Infosphere.

[adritheonly@mastodon.social]

@kmeisthax @overholt Yes, that makes sense. When I moved to a really small place and did not have space for my library of just under 3000 books, my only solution was to scan them but also the destructive way since that is the only scanner I had access to.

The issue here is not destructive scanning but what Anthropic intends with the end result. Knowledge monopoly via the internet? Warping knowledge to suit its makers view of the world? Difficult to do. There will always be my library.

[izzaboo@tilde.zone]

@kagan @overholt Also, when I look at a book there is *actual content* but that is only part of, and different than, my *actual experience* of looking at a book.

[steveclough@metalhead.club]

@overholt I mean, no, no no. Stupid. As expected, utterly missing the point of books.

Hardly surprising given that they are only interested in words that sound like they make sense.

[emassey0135@caneandable.social]

@overholt @jpellis2008 When you train an AI model on information, it is like giving it to a lossy compression algorithm, so information can never be fully preserved that way. And its not like MP3 or a copy machine either, where what is lost is just characteristics of the information not the information itself. The information itself is subject to lossy compression, since the AI could give you things like summaries of the book and its sections, its writing style, quotes of some passages, etc but it wouldn't be able to quote the whole book back to you. Also it is doubtful that AI models created inside companies like that would last longer than a physical book or a digital archival copy of it, meaning even the lossily compressed copy created this way is temporary.

[r_alb@mastodon.social]

@shadows
Had very similar thoughts when I read the quote. Those insufferable bros...

@librarymonster @overholt

[metaning@mastodon.social]

@overholt I would feed the leaders of ISBNdb alive into a wood chipper, and feel no remorse.

[modernmaslow@techhub.social]

@overholt Soon the only bind we'll know, is the one AI put us in.

[ngrn@soc.ngrn.uk]

@overholt It blows my mind that these idiots are so far detached from reality that destroying books to them is just an optics issue.