Library of Alexandria 0.2: AI Companies Are Buying Rare Books in Bulk and Destroying Them After Training Their AI Models

History remembers civilizations not by how rich people were, but by what they thought, learnt, invented, and discovered. 

The Library of Alexandria still feels heavy as one of history’s biggest tragedies, not because the building got burned, but because humanity lost whole bodies of knowledge that we couldn’t just recover later. Wars wrecked manuscripts, dictators put a chokehold on books, and time itself erased countless stories, bit by bit, until they were gone. Each generation looks back at those disappearances with this regret, wondering how history might have moved in a different direction if those volumes had survived.

But here we are, in one of the most technologically advanced eras humans have ever lived through, and it seems we’re making a similar mistake again, not with fire, but with scanners this time. And who is to be blamed? Human greed. 

Videos of rare books being cut are circulating online. Fresh reports say AI companies have been buying massive quantities of physical books, even rare copies and out-of-print editions, then cutting off their spines and scanning every page just to train artificial intelligence models. Afterward, the original books get destroyed, because it’s supposedly cheaper and quicker than preserving them intact. Court papers and legal hearings indicate that this kind of destructive scanning is now baked into the AI industry’s competitive push for high-quality, human-written data.

Legally, a lot of this might be allowed. Morally, though, it leaves behind questions no courtroom can really settle , at least not in any satisfying way. Because a book isn’t only information written on papers. A book is memory , culture, and history you can actually hold in your hands.

Destroying Books Was a Choice, Not a Necessity

The most uncomfortable aspect of this story is that the destruction of books was never an unavoidable consequence of digitisation. And that's where this approach becomes ethically and morally wrong.  Libraries, museums and universities have spent decades preserving fragile manuscripts while simultaneously making them available digitally. Many institutions, including Harvard University, have shown that we can scan and digitalise books without actually destroying the original!

However special equipment, trained archivists, careful handling and much more time is needed for preservation of books, and this is the economical factor that is being avoided. The same operation of cutting the spine off a book and putting it through an industrial scanner is a dramatic improvement in both speed and cost. The financial reward for the industry that's racing to develop the next generation of AI models is clear.

This is where the discussion goes beyond the technical and into the philosophical realm. But whether books can be preserved is no longer the question. They can. The problem is that these big AI Companies who can easily pay for preserving these books are okay with destroying them.

No AI company has publicly claimed that history is expendable, nor is there evidence that they intend to erase literary heritage. But, every industry shows its priorities in the decisions it takes. With the continuing triumph of speed over preservation, when it's done for efficiency and cost-saving, a question arises: Are the commercial considerations starting to undermine cultural responsibility?

Throughout the history of technology, new innovations have always promised to make life simpler. It should never lead to us being lax in what we inherit. 

A Book Is More Than Just The Words Inside It Is

There are people who believe in destructive scanning who say that nothing is being destroyed since everything is kept in a digital format. If the information persists, why should the material object? The answer is no big surprise for those who have a deep passion for books.

A book is not a piece of information. If it were just information, printed books would have been eliminated on the day the Internet came along. After all, almost all classic novels, scientific papers and historical texts are on the internet and available within a few seconds. But bookstores are still thriving, libraries are never empty and millions of readers opt to read books instead of e-books.

This decision does not have much to do with convenience. Books are experiences. They smell of old paper, feel like worn paper and have the feel of a product designed to last generations. The second hand book will most likely be marked in some way with underlined passages, missing bookmarks or handwritten notes that were written by an owner who the reader will never see. Those little flaws become the unspoken dialogue in time.

The words narrate one story while the book tells another. This is something that people who don't read novels very often do know unconsciously. Often students will print out key notes rather than read the notes on screen. Even if you've a laptop and a tablet, writers still carry notebooks around. Individuals write letters rather than scan them into folders. Childhood diaries, family recipe books and old journals are kept as they have emotional value in themselves.

Humans aren't just focused on consuming information, we establish connections with it. What might be preserved here is the text, but all the things that made books part of our common literary culture are lost.

Profit Is Starting To Decide What We Preserve

Books are necessary for AI companies because they require humanity. But guess the irony is greater.  Technology has been offering a vision of a future where machines would produce unlimited knowledge for years. Rather, the AI craze has exposed something new. Human creativity is used in machines. They need authentic writing produced by authors, researchers, journalists and thinkers long before artificial intelligence became part of everyday life. 

That reliance has made books one of the most valuable commodities in the AI economy. The use of AI to learn from human knowledge isn't inherently bad. All generations have brought learning from others. What concerns them now is that centuries of human creativity are becoming seen as just a source for commercial products.

A rare book that used to be cultural heritage gets turned into training data, a cultural heritage book becomes a training data set, a book that was exclusively written, becomes a mere data, and the books that survived invasions, world wars and natural calamities are all now a mere set of data.  The transformation is subtle, but its implications are profound. History stops being something to preserve and gradually becomes something to process.

This may be the natural consequence of a competitive industry. All firms wish for bigger data sets, improved models, and quicker product releases. Growth is expected by investors and timing is all about money and efficiency in the markets. In such an environment, it's logical for a company to select the less expensive choice. However, economically sound is not necessarily ethically sound, isn’t it?

Civilisations have been able to thrive because previous generations saved far more than they immediately needed. . They preserved monuments for their own sake, defended languages known by only a few, and preserved libraries containing books which many never read. They knew that culture does not necessarily have to be quantified monetarily.

Today It Is Rare Books. Where Does the Line Move?

At present, reports largely involve rare and out-of-print books rather than unique historical manuscripts. It would be inaccurate to suggest otherwise. But technological history teaches us that practices rarely remain confined to their original boundaries.

The original social media were never designed to affect elections. Smartphones weren't meant to be a constant presence in every human moment. Streaming services had no idea that they would come along and dramatically change entire film industries. Technology grows where it is encouraged to grow. The same applies with this also.

With the growing diversity and realness of the sources that AI models depend on for human writing, society will have more challenging questions to answer. What happens when there is very little regional literature and it becomes very profitable? Where do endangered-language books go? If there are only a few physical copies remaining, what then?

These are not forecasts or estimates, but questions that deserve to be asked before circumstances force us to answer them. Once destroyed, there is no second chance for reconsidering a rare book. A copy in digital form can retain all words but can it maintain authenticity?

In Trying to Make AI More Human, Are We Making Humans More Like Machines?

It might be the strangest irony of this technological revolution that it has nothing to do with artificial intelligence. It is all about us!

Over the past few decades, researchers have been making efforts to equip machines with human-like thinking capabilities. They have taught them to recognise emotions and communicate naturally, imitate creativity, and understand language. Each advancement has led to machines being closer and closer to human actions. But, as machines grow increasingly human, human beings appear to be going in the other direction.

We are becoming more and more about celebrating speed over patience, summaries over understanding and efficiency over experience. AI composes our messages, condenses our books, creates our reports and spawns our concepts. We skim, rather than read. Rather than reflect, we encourage. Rather than make, we automate.

The threat is not one of total replacement of the human race by AI. The greatest risk is that man slowly starts to imbibe the values of machines.  When every choice is made on the basis of speed, cost and optimisation, we may end up being a civilization of output over understanding. One day we might know that we have the most intelligent machines ever created and forget to remember why books, art and history were important other than for the information they contained.

The Future Should Not Make Us Forget The Past

AI is definitely one of the greatest technological accomplishments of humankind. It could change medicine, education, scientific discovery and many other things. It should keep on going. 

True progress never came when cultural responsibility was given up. It is not a debate on scanning books or training AI Models. It's about the values that will define the future that we are creating. When it comes to preserving our literary heritage, if we're only looking at reducing expenses, we might end up with a world that's been digitised, but with far fewer books remaining.

Inventions are not the only things by which a civilisation is judged in history but also the treasures it chooses to preserve. Maybe someday AI will have the ability to recall all the books people have ever written. But what’s a bigger issue is whether humans will still be able to recall why those books even existed. The problem is trivial right now but in the long run it’ll change the world, and that, perhaps, won’t be for good.