IT’S a little disturbing to live long enough to see fictional dystopias become reality. Take Ray Bradbury’s Fahrenheit 451 — published in 1953 — which depicts a world in which books are outlawed and specially designated ‘firemen’ are tasked with destroying these offending objects wherever they are found.
The soulless world imagined by Bradbury didn’t take shape overnight; as society became more and more fast-paced, people began to prefer their entertainment to be as shallow as possible, which led to wall-sized interactive televisions in every home constantly broadcasting bright and chaotic programming for people to numb themselves with. People wore tiny wireless ‘seashell’ earbuds streaming a constant “electronic ocean of sound” into their ears, even while asleep, effectively drowning out the world and any need for actual human interaction. What social activities remained were vapid and violent; there were places where you could go and engage in wanton destruction with full social sanction.
As for schools, they were now meant only to produce “runner, jumpers [and] racers” and not thinkers. Naturally, subjects like history and philosophy were no longer needed. Independent thought was tedious and troubling and so was done away with.
The best part? None of this was the result of some Big Brother State intent of anaesthetising the public; the process began from the bottom-up, with society itself demanding that literature be condensed, with classics first being reduced to digests, and then to summaries, and then finally being legislated out of existence. Seventy-three years later, you yourself can judge how much of Bradbury’s nightmare has come to pass, albeit in a somewhat altered form.
Recently we learned that AI companies have been bulk-buying thousands upon thousands of old and rare books — sometimes the only remaining copies — to be scanned and fed into their machines and then shredded and destroyed because that is apparently the most efficient method available. Why old books? Because the Large Language Models have already scraped the entire internet and are now at risk of ‘model collapse’ where, because AI is increasingly relying on material produced by other AI, the quality of output is degrading. It’s like making photocopies of photocopies until the text is no longer legible.
Why search page after page when summaries are available?
Hence, they must turn to older, human-written texts from an age long before AI or even the internet that provide the kind of rigorous high-quality knowledge and writing that is hard to come by in these times. Once you’ve absorbed the irony of AI considering AI-produced writing as sub-par, consider the wider implications: The irreplaceable knowledge contained in these soon-to-be shredded books would now be the sole property of companies like Anthropic, whose intent it is to monopolise that knowledge and sell it to you on a meter with no way for you to discern whether the output of the LLMs you will come to rely on is factual or simply what those who write — or manipulate — the algorithms want you to believe.
In that latter category falls Israel, which created a fake thinktank called the ‘Hanover institute for Public Policy’. It has no physical address or staff and its reports carry no bylines but it has nevertheless produced 124 “academic-looking reports” comprising of over 560,000 words of pro-Israeli content in just nine days, none of which are meant for direct human consumption but for LLMs to pick up and reproduce. With headlines deliberately resembling the kind of queries one asks of AI chatbots, like ‘is Israel carrying out a genocide in Gaza?’ or ‘has Israel driven Palestinians from their homes?’ these ‘reports’ are meant to be consumed by AI and then vo-mited out in response to user que-ries.
Do a Google search right now and the search engine will not immediately take you to a list of websites. Instead, an AI summary will pop up on your screen. It’s convenient and often useful of course, but this new feature — one that we did not choose — has also reduced organic website traffic by anywhere from 38 per cent to 60pc. This means that smaller, but highly useful independent news publishers and information providers are seeing less traffic and thus less income and may soon cease to exist. The bigger fish in the media ecosystem can take the hit, and we already know how easy it is for the powers that be to manage those bigger fish.
So why go through the trouble of searching page after page (and maybe learning something new in the process, God forbid) when easily digestible summaries are a click away? Why sit through a nuanced, perhaps hours-long lecture on YouTube when you can just read the AI summary? Why take the time and effort to think when someone else can do it for you? Why bother reading any of this when you can just watch reels?
The writer is a journalist.
Published in Dawn, September 7th, 2026
