[ad_1]
Your favourite rare book may disappear someday, coming not from a certain established fact or the consideration of future evolution, but it is a concern that is increasingly being discussed as AI companies race to find more human-written material to train their models. In that process, millions of books are being scanned and, in some cases, physically destroyed after digitisation.
Earlier this year, The Washington Post reported on Anthropic’s internal initiative, called Project Panama, under which the AI company purchased millions of physical books, scanned them and discarded the originals after the process.
“Project Panama is our effort to destructively scan all the books in the world. We don’t want it to be known that we are working on this,” one internal planning document stated, according to The Washington Post.
The question of originality in books, the chances of manipulation after scanning, who gets access to the original source, and the commercialisation of such material all come into play.
Preserving and digitising knowledge has been widely discussed over the past few years. India has taken a significant step by digitising many manuscripts and making them accessible to millions.
The debate, however, extends beyond the physical destruction of books.
A GROWING CONCERN AMONG BOOK COLLECTORS
The developments have also drawn attention in India, where collectors and booksellers dealing with antiquarian books say preserving original copies remains important even as digitisation expands.
India Today Digital spoke to Rishabh Jain, who runs Memoirs of India, one of Delhi’s oldest antiquarian book businesses.
Jain believes digitisation can preserve text, but not every aspect of a historical book.
“There are many books with illustrations having hand-painted gold. Old-school techniques like lithography and aquatints were labour and skill-intensive methods that produced limited prints. If digitisation had the same feeling and value as the originals, the whole art industry would collapse,” he said.
According to him, removing a book’s spine should only happen when restoration is no longer possible.
“It is unfortunate and heartbreaking to destroy anything of original heritage. There are other ways to digitise a book without damaging its originality,” he added.
THE LOSS GOES BEYOND PAPER
For collectors, the concern is not limited to the physical object. Many rare books are primary historical sources written by eyewitnesses, historians or scholars who documented events using the earliest available records.
Jain says AI systems still depend on original material for reliable information.
“AI data could be true or false. To verify information, we need the original source. From cave paintings to manuscripts and books, these are the earliest documented records. AI can interpret history only if those sources are available for digitisation. Since we are still far from digitising everything, the available source material remains limited,” he explained.
He also pointed to another challenge in the age of AI-generated content.
“We now live in an era where anyone can publish their own interpretation of history. That can result in contradictory versions of the same event. The original source becomes even more important.”
SCARCITY IS ALREADY VISIBLE
While AI companies have not directly approached his business to buy books in bulk, Jain says rare books are becoming harder to find.
“There is definitely a rise in scarcity. Prices in our industry depend on demand and supply. The supply hasn’t been generous while demand has increased, so prices have gone up significantly. We don’t see them coming down anytime soon.”
Interestingly, he says younger readers are also showing renewed interest in original editions from the 18th and 19th centuries, moving away from digital reading devices and returning to printed books.
The issue has also sparked discussion online. On Reddit, several users questioned whether destroying books can be described as preservation.
One user wrote that less popular books are often printed in much smaller numbers than bestsellers, making every surviving copy valuable and increasing the risk of permanent loss.
The debate is no longer only about AI or copyright. It is also about what happens to books once they become training data, and whether preserving knowledge should mean losing the original copy.
Many readers have raised the same concern online. One Reddit user summed it up by saying, “They’re not preserving. They’re destroying the books and the contents are just being used as training data.”
India has taken a different route. Through the National Mission for Manuscripts (NMM) and digitisation projects by institutions such as the Indira Gandhi National Centre for the Arts (IGNCA), lakhs of rare manuscripts have been digitised while the original documents continue to be preserved.
Anthropic says its process is legal, and a US court ruled that using legally purchased books to train AI qualified as “fair use” because it was transformative. The company, however, still faces copyright claims over the alleged use of pirated books.
As AI companies continue looking for more books to train their models, one question remains: can technology learn from the past without leaving the original books behind?
– Ends
[ad_2]
Source link