News AI companies are buying up antique books in bulk, scanning them and then destroying (1 Viewer)

eclipsoul said:

we all died in 2020 and this is hell ​

Click to expand...

RandAlThor said:

Its my understanding this is not the case.

It was reported as Anthropic, using some of the old goggle ppl from the massive project to digitalize all the books years back.
while i do not love the companies doing this, reporting seems to have become dramatized.
the fair use battles in the court from googles project led them here.
The gist of all the ruling's were that fair use allowed for A copy in an owners hands so if they wanted to scan it. the physical book must be destroyed.

This is far different than the google project to make an searchable /accessible database where they went to extreme lengths to preserve teh books.
then they are using that to train their models only.
They are buying books under copyright ie post 1930's. That are low count, think something on DOS tables ect.


Historic books in the public domain have long been read in.

Now some companies ie Xai have are going to single page scan low count books.

Also i would expect the Library of congress has a copy of all these.

Click to expand...

LoveYooShaSha said:

Late but for last point: so did (supposedly) Library of Alexandria hv a copy of many manuscripts/papyrii but look wat happened there!

if u watched MI:8 or maybe other shows, theres a repository of seeds/dna (digitised maybe) kept hidden in case of apocalyptic collapse see maybe Canticle for Leibovitz(no repository) or Asimov's Foundation(Galactic repository/encyclopedia buy with a twist) excluding the fact its fiction, I don't understand/see how it could work, tech to revive/usefully produce sthg from seeds would fail with fall of civilisation.

so my point is Library of Congress might be as susceptible as Library of Alexrandria by which i mean a single point of failure might result in books being lost forever. If there were multiple books in different o
Places/repositories where they were preserved but tats not wat these destructive behemoth companies plan , rather destroy n survive n plunder wat is left seems to be the plan. Sigh

Click to expand...

RandAlThor said:

The problem is. There Has to be multiple Digital back ups, but the copyright holders refuse to let there be a useful repository.
End of the day every thing will be in all these models. Non 'western' companies / governments will not give a rip about copyright. They should have let google make the PDF data base and taken the minimal payment then.
Now they got what 3 k per book from anthropic and not much else will be comming.

Click to expand...