Aardvark Daily

Please visit the sponsor!

The importance of hardcopy

Yesterday I received an email from a reader.

They pointed me to a story highlighting the fact that AI companies have been buying books and scanning them so that the information contained within their covers can train their systems.

I guess this is one way of trying to avoid copyright claims from publishers griping that their content has been scraped from the Net without permission. Is buying a book, scanning it, then training your AI on its contents really any different to a person buying a book, reading it and learning from what they read?

The writer of yesterday's email raises some very interesting points however.

Books (hardcopy) have fantastic durability.

Long after some distopian disaster sends us all back to the stone age and turns our bewildering array of complex electronics into useless paperweights, books will still work and they'll actually become our only source of archived knowledge.

Even the custom of handing down knowledge and history by word of mouth from generation to generation is flawed in much the same way that repeatedly taking a photocopy of a photocopy soon produces a result that bares little resemblence to the original. Slight changes made at each batton change soon distort and corrupt the facts.

That really only leaves us with inky stains on the pulped remains of dead trees as a robust, reliable, non-tech-reliant medium for transferring essential knowledge.

Another benefit of books is that they are effectively immutable.

Books, once printed, can not easily be changed to suit the fashions, politics or beliefs of the new day. They allow us to freeze the facts in time and render them immune to "shaping" by external influences.

This is the exact opposite of our electronic, cloud-based storage systems.

Imagine if you will, the government of the day decides that, because they're thinking of the children, all references to Winnie the Pooh be eradicated from the world. If you've got one of those old books from the 1960s then Pooh will live on forever but if all your books are ebooks on connected devices then they can be edited, re-worded or even just totally erased in the blink of an eye.

Without your permission or knowledge -- Pooh disappears from the contemprary world and within a generation or two, nobody left alive even remembers he existed.

This makes it incredibly easy for politicians and others to rewrite history to suit their own ends.

Books are an essential and pretty reliable link to the past. A bunch of bytes on a hard drive in a datacentre somewhere is not.

The author of the email also reminded me of the joys that come from just casually browsing through my pile of Popular Science and Popular Mechanics magazines from as far back as the early 1940s. These things are glorious for so many reasons.

They provide an incredible window into the culture of the era (ads for cigarettes, women being depicted as always holding a vacuum cleaner or frying pan) as well as the technology and the structure of society. They are truly historical documents.

In the second quarter of the third millennium, I suspect that very few of us have subscriptions to printed periodicals. Instead we visit websites, listen to podcasts or cast our gaze to the occasional YouTube video. Where are the guarantees that this content will be archived for future generations to enjoy -- even if we don't suffer some kind of tech-apocalypse in the meantime?

Let's also not forget that all these books (and other stuff) is now being used to train massive AI systems so I expect that at some stage, we'll simply ignore actual historical documents and instead rely on those AI systems to tell us of the past. We're seeing this happen already with search results on Google. When you do a search, most people just read the AI summary and don't even bother to follow the links it contains or the *actual* search results that follow.

As this trend continues, our view of the world, past and present, becomes fully shaped by those AI systems. Pretty soon Google may not even bother showing links to websites -- why waste the bandwidth when nobody uses them anyway?

What happens when AI decides that there was no holocaust and those who remember what happened have all died off, leaving nobody to refute the AI claims? Does history get re-written by chunks of silicon in a datacentre? Could the past be changed by all-powerful "big tech" or overbearing authoritarian governments to suit their own ends?

If we don't retain permanent, immutable records in the form of books, the re-writing of our history becomes an increasingly likely scenario.

To be honest, I can see the day when books become illegal to posess because of the threat they pose to those who would seek to control a narrative and rewrite history.

All the more reason to keep, protect and revere them.

Carpe Diem folks!

Please visit the sponsor!

Here is a PERMANENT link to this column

Rank This Aardvark Page

Change Font