AI companies are reportedly buying physical books, cutting away their bindings, scanning their pages, and destroying the originals after their contents have been extracted. Rare and out-of-print works may be disappearing in the process. Futurism, July 25, 2026.
The immediate concern is the loss of books. But the deeper issue is the logic that makes such destruction appear reasonable.
Within that logic, a book is merely a container. Its text is valuable because it can improve a model; the object itself is valuable only until its information has been captured. Once the words have been converted into data, the source becomes waste.
This is not an accidental side effect of artificial intelligence. It is the consequence of an extractive way of understanding knowledge.
Culture is approached as a resource to be acquired, processed, and absorbed. Meaning is separated from origin. Knowledge is detached from the people, places, and relationships that produced it. The resulting system may contain an extraordinary quantity of human expression while remaining disconnected from the human worlds in which that expression has meaning.
A book is more than its text. Its language, edition, typography, materials, annotations, circulation, ownership, and physical condition are also forms of knowledge. The object carries evidence of how ideas moved through time and between people.
The same is true of culture more broadly.
A song is not only an audio file. A ceremony is not only a sequence of movements. A language is not only a dataset. An archive is not simply a supply of documents. Each exists within a living network of memory, identity, authority, interpretation, and participation.
Digitizing culture does not automatically protect that network. A perfect scan can preserve an image while severing its meaning. A model can reproduce a cultural form without understanding its context. A database can contain millions of objects while making their creators and custodians effectively invisible.
Preservation is essential, but it is only the beginning.
The larger task is to keep cultural knowledge connected, to its source, its history, its relationships, its custodians, and its continuing life. Culture survives through use, interpretation, transmission, disagreement, reinvention, and encounter. It must be carried forward, not merely stored.
Artificial intelligence could help make those connections visible. It could allow people to move across languages, collections, disciplines, and national boundaries. It could reveal relationships that no single archive or institution can see alone. It could reconnect communities with dispersed histories and open cultural knowledge to new forms of education, research, and creation.
But that possibility requires a different foundation.
It requires systems that preserve provenance rather than erase it. Systems that distinguish access from ownership and knowledge from permission. Systems that recognize communities as participants, not simply sources of material. Systems that return value to the cultural ecosystems from which intelligence is drawn.
The central question is therefore not whether machines can ingest culture.
They clearly can.
The question is what remains after they do.
Does the source remain intact? Is its origin visible? Can people find their way back to the work, community, language, or institution behind the output? Does technological use strengthen the continued life of the culture, or merely transfer its value elsewhere?
The future of AI cannot be measured only by how much human knowledge a model can absorb. It must also be measured by what the system enables humanity to retain, understand, connect, and create.
The real alternative to extraction is not passive preservation. It is cultural continuity: a living relationship between memory and possibility, inheritance and invention, the people who came before and those still to come.
The future cannot be built by consuming the past and discarding its remains.
It must be built by expanding the life of what we inherit.