AI boom sparks rush to buy and scan old books

In an otherwise quiet summer season when news is sparse, the literary world erupted with concerns about what was happening to used books, even some single-edition rare ones.
The story itself is a cloak-and-dagger one.
Booksellers in Britain and the US noted a spike in sales of various books in their inventory which had not moved for years. Suddenly, sales rose dramatically, but it was not clear who the buyers were.
Further research showed that they were going into warehouses associated with AI companies.
The way that AI software uses real-world data to train itself has already been discussed. Tailors, for example, are filmed wearing cameras on their foreheads so that the complete process of cutting and stitching fabric from scratch is recorded for AI use at a later stage.
Books are making a similar journey – individual books are being bought, torn from their spines and scanned at an industrial scale. This is, of course, to make it possible for the software to digest huge amounts of data in the form of language, information, even plot and character intricacies from literature written hundreds of years ago.
The AI would then categorise such data for future consumption.
Evidence for this emerged through an investigation by journalist Emanuel Maiberg, who inserted an electronic tag into an order of books and found that the box had landed at an Amazon warehouse.
The explanation for this industrial-scale scanning of old books, which are only available in print form, is quite simple. AI has already used all the digitally available data.
According to those who work in these areas, this leads to 'model collapse', meaning that AI training on repetitive existing data will make the product worse, not better than what has so far existed.
AI thus needs more content. And what is available is the printed word – books going back centuries, in different languages. Each book is unique, talks to a distinct time and place, to a unique culture and a specific audience.
AI seeks to bring it all together under one umbrella – or one computer tab.
In fact, the rush is now to acquire and scan books published before 2022 to ensure that there is no AI interference in those books.
Because this is a new phenomenon, the ethics around such activity are unclear. There are copyright laws that could be infringed with living authors, or those with functioning estates which are still in control of the original material. But that is not true for texts written 200 years ago.
Technically and legally, once a book is bought, the owner can do anything with it, including tearing it apart and later shredding it. This is exactly what seems to be happening to physical books. But the larger questions are only just beginning to be asked.
For example, if all literature is now ‘owned’ by a large private multinational company, what do human thought, originality and creativity actually mean? If knowledge is under the control of the few, all writing will ultimately be part of a larger muddle of words with nothing unique to merit it.
Literary writing has also become a commodity, in this case. The individuality of writers will be erased if AI models learn from their stories and styles and churn out dozens of books within the time that it takes to write one thoughtful book. This will then lead to serious trust issues as readers no longer know what is human and what is machine-made.
There could be solutions to these immediate and long-term concerns. But those would only be possible if books are still being valued for themselves and not as a commodity to be digitalised.