Algorithm and the Fading Physical Memory of Civilisation

Reports of the destruction of rare manuscripts and ancient books to train artificial intelligence (AI) models mark a deeply troubling new chapter in the long history of humanity’s relationship with its technology. Behind the grand narrative of leaps in artificial intelligence progress, a structured and systematic process of cultural cannibalism and epistemicide is underway.

In pursuit of high-speed raw data volume, centuries-old leather-bound volumes are forcibly dismantled, book spines cut without mercy, and historic paper sheets separated to the point of destruction so they can be fed into automated scanning machines. This act is a true reflection of radical reductionism, where all human knowledge is squeezed into mere lines of digital text, while the physical body in which that knowledge took root is sacrificed without regret.

The root of this cultural crisis lies in the technology industry’s ontological error in viewing the essence of a book or ancient manuscript. For developers of large language models, an ancient manuscript is merely a passive container of text, a string of characters to be extracted word by word to increase the volume of training tokens.

Yet ontologically, a rare manuscript is a holistic, unique, and multidimensional historical artefact. The texture of handmade paper, the composition of binding materials, the blend of ink types, and the marginalia left by readers across the ages provide invaluable empirical evidence of chemical technology, political economy, and the dynamics of thought in the era of its creation.

When technology corporations destroy the physical form of a book to accelerate data extraction, they are effectively severing the authentic historical link between the past and the future. What remains is a digital shadow of binary code that is highly vulnerable to format erosion, storage media extinction, algorithmic manipulation, and unilateral control by commercial platform owners.

From an epistemological perspective, separating text from its physical material through a destructive mechanical process creates a serious threat of systemic hallucination and loss of authentic context. The text absorbed by AI is extracted sterilely, without the physical and historical context in which it was embedded. Once the physical object is destroyed, future scientific communities will no longer have physical ground truth to re-verify the authenticity of the manuscript. Human knowledge suddenly becomes entirely dependent on a digital version locked inside the black box of a technology corporation.

This phenomenon is deeply rooted in modern extractive capitalism and the growing fetishism of data. Just as the Industrial Revolution exploited natural resources without regard for environmental recovery, the AI industry is now commodifying and enclosing cultural heritage.

Ancient manuscripts that are in fact valuable public property are destroyed to be fed into closed, paid commercial algorithmic models. The claim that this destructive scanning is a form of ‘information democratisation’ is a profoundly misguided and misleading rationalisation. The gain of a few weeks in extraction speed is in no way comparable to the permanent extinction of historical artefacts that have survived multiple civilisations for hundreds of years.

It is a bitter paradox that AI, developed with the ambition of mimicking human intelligence, instead displays barbaric behaviour by devouring and destroying the roots of its own history. This is a form of sophisticated barbarism, where mechanical efficiency and speed are worshipped above respect for the memory and soul of civilisation. This destruction not only harms academic disciplines such as codicology and philology, but also separates future generations from the profound sensory experience of directly encountering the physical traces of the past.

To halt this cultural cannibalism, the academic community, researchers, and international cultural institutions must take a firm stance immediately. An international moratorium on destructive scanning methods is needed, accompanied by strict data ethics regulations to ensure that technological development is not paid for with the annihilation of cultural heritage.

The AI industry must also be required to invest its resources in high-precision, non-destructive robotic scanning technology. Technology should exist to care for and extend the life of humanity’s collective memory, not to become a tool of physical elimination that legitimises destruction. If humanity allows its civilisation to be built upon the rubble of artefacts destroyed by its own machines, then the technological progress we boast of today is in truth merely a new form of ignorance wrapped in great sophistication.