Oh right. But anyway, nobody knows what needs preservation, it's a basic problem of life, somebody usually mentions the BBC throwing out boring old Doctor Who tapes to save archive space because nobody liked it any more at that point in time. Some things should probably be thrown out now and then, I suppose.
The alternative for many of these un(der)appreciated books is that they will get unceremoniously dumped in the future anyway. The publishing industry and libraries etc dispose off lots and lots of books.
So at least with the AI companies they are scanning them and preserving them digitally. Not just in the trained weights, but also as raw training data for future runs.
P.S. I'm not sure why you need to make fun of your own ignorance? Just look up the word you don't know and don't mention it?
The AI companies’ working assumption is that if someone found it worth printing, it has enough information content to help train a model. That assumption might be invalid with some of the more rambling self-published books, however.
You are giving me too much credit: it's made up evidence. It's an illustration that rarity doesn't equal value, and doesn't depend on whether I actually own a wastebook or ten.
Your point is understood but the crux of the issue is a bit similar to capital punishment, the argument is that risk of losing even one innocent person or useful book isn't worth taking; specially considering the value created for society as part of such risky undertaking, whatever it is exercising capital punishment or scanning and destroying books.
Whenever you build a highway or a bridge or a power plant, your engineers have to put a money value on human life, or at least the worth of a statistical human life. Just to make ordinary engineering decisions.
And refusing to do this exercise just means that you behave as-if you put a really silly number on the value of human life, and probably not consistent between different parts of the project.
So I don't quite agree with these taboos in the absolute.
(I'm still against capital punishment on practical grounds.)
For books it's similar: if you taboo book destruction for the AI training folks, that doesn't rescue books from their ordinary pre-AI life cycle of getting destroyed all the time in the course of running a publisher or a library or a second-hand book store.
In fact, the AI craze is what's giving rare books _value_ and incentivises people to dig them up and preserve them. Or at least preserve them long enough to be scanned.
The scanning might destroy the physical copy of that book, but they save the contents. That's the whole point of scanning after all.
I am well aware of the Statistical Value of Life and how it impacts projects. But that is generally used for allocating preventative measures. No road is going to get approved if it is going to result in random deaths by design.
Like I said, at this scale, there are no guarantees for anything. Very likely will there be a unique copy of an invaluable book or letter an the person feeding it to the scanner will not know and the book get destroyed.
Like did Icelandic author Þórbergur Þórðarson ever write an a book about Esperanto, and send the only copy of it to Halldór Laxness when he was in Los Angeles? I don‘t know, but it is certainly something he is likely to have done. If such a book exists it would be invaluable to both Icelandic culture and to Esprentists. It likely would have stayed in Los Angeles where nobody would know the significance of it until it ended up in an estate sale, a used book store, and then finally destroyed by an AI company never to be discovered.
My hypothetical is just one of trillions of possibilities. At this scale very likely several of these possibilities will unessiseraly remain unknown unknowns forever.
Well, at least afterwards the book is scanned and preserved digitally in their archives of training data.
If the book was just rotting away in some forgotten bookstore, it would more likely be unceremoniously disposed off in the future without anyone scanning it first.
Your what now? Made from your personal waste? That does sound unique.
https://en.wiktionary.org/wiki/wastebook
Oh right. But anyway, nobody knows what needs preservation, it's a basic problem of life, somebody usually mentions the BBC throwing out boring old Doctor Who tapes to save archive space because nobody liked it any more at that point in time. Some things should probably be thrown out now and then, I suppose.
The alternative for many of these un(der)appreciated books is that they will get unceremoniously dumped in the future anyway. The publishing industry and libraries etc dispose off lots and lots of books.
So at least with the AI companies they are scanning them and preserving them digitally. Not just in the trained weights, but also as raw training data for future runs.
P.S. I'm not sure why you need to make fun of your own ignorance? Just look up the word you don't know and don't mention it?
The AI companies’ working assumption is that if someone found it worth printing, it has enough information content to help train a model. That assumption might be invalid with some of the more rambling self-published books, however.
https://en.wikipedia.org/wiki/Anecdotal_evidence
You are giving me too much credit: it's made up evidence. It's an illustration that rarity doesn't equal value, and doesn't depend on whether I actually own a wastebook or ten.
Your point is understood but the crux of the issue is a bit similar to capital punishment, the argument is that risk of losing even one innocent person or useful book isn't worth taking; specially considering the value created for society as part of such risky undertaking, whatever it is exercising capital punishment or scanning and destroying books.
Whenever you build a highway or a bridge or a power plant, your engineers have to put a money value on human life, or at least the worth of a statistical human life. Just to make ordinary engineering decisions.
And refusing to do this exercise just means that you behave as-if you put a really silly number on the value of human life, and probably not consistent between different parts of the project.
So I don't quite agree with these taboos in the absolute.
(I'm still against capital punishment on practical grounds.)
For books it's similar: if you taboo book destruction for the AI training folks, that doesn't rescue books from their ordinary pre-AI life cycle of getting destroyed all the time in the course of running a publisher or a library or a second-hand book store.
In fact, the AI craze is what's giving rare books _value_ and incentivises people to dig them up and preserve them. Or at least preserve them long enough to be scanned.
The scanning might destroy the physical copy of that book, but they save the contents. That's the whole point of scanning after all.
I am well aware of the Statistical Value of Life and how it impacts projects. But that is generally used for allocating preventative measures. No road is going to get approved if it is going to result in random deaths by design.
Roads with cars on it do result in random deaths. The preventative measures only prevent deaths that the road itself caused.
Like I said, at this scale, there are no guarantees for anything. Very likely will there be a unique copy of an invaluable book or letter an the person feeding it to the scanner will not know and the book get destroyed.
Like did Icelandic author Þórbergur Þórðarson ever write an a book about Esperanto, and send the only copy of it to Halldór Laxness when he was in Los Angeles? I don‘t know, but it is certainly something he is likely to have done. If such a book exists it would be invaluable to both Icelandic culture and to Esprentists. It likely would have stayed in Los Angeles where nobody would know the significance of it until it ended up in an estate sale, a used book store, and then finally destroyed by an AI company never to be discovered.
My hypothetical is just one of trillions of possibilities. At this scale very likely several of these possibilities will unessiseraly remain unknown unknowns forever.
Well, at least afterwards the book is scanned and preserved digitally in their archives of training data.
If the book was just rotting away in some forgotten bookstore, it would more likely be unceremoniously disposed off in the future without anyone scanning it first.