One of the great writers of the twentieth century prefaced, 84 years ago, a paradox now threatening the Internet
How will the internet and human society be inextricably related to it will evolve in the following decades? Fiction writers have explored some possibilities, the magazine shows The Conversation.
In “Fall”, his 2019 novel, the author of Science-Fiction Neal Stephenson imagined a near future in which the Internet still exists. But it is so polluted with misinformation, false information and advertisements that it has become largely unusable.
This is what we see now on social networks such as Facebook, Tiktok or “X”, which have become true misinformation relays or distribution channels of ads that promote scams.
The history of Stephenson as a visionary is impressive-he anticipated the metal in his 1992 novel, “Snow Crash”, and a key element in “The Diamond Age” (1995) is an interactive manual that works in a similar way.
At first glance, more popular chatbots in the real world seem to offer a solution to the misinformation epidemic. Presenting factual content, they could be alternative sources of quality information, quickly providing information for users who ask them questions.
The irony is that, in fact, precisely the information generated by the chats can be the greatest danger to the future of the Internet – an anticipated danger ago by the Argentine writer Jorge Luis Borges, one of the literary titans of the 20th century.
The ascension of the chatbotes
Today, a significant part of the Internet is still made up of factual and, apparently, truthful-articles and materials that have been revised by experts, verified or validated in a way.
Developers of large linguistic models (LLM) – the engines behind chats such as Chatgpt, Childhood, Claude or Gamini – have taken full advantage of this resource.
In order to achieve their “magic”, these models must be trained on huge amounts of quality text. Much of this content has already been extracted from online sources and introduced into LLM models when they were at the beginning of the road.
The problem is that the Internet, no matter how vast it may seem, is a finite resource. The quality texts that have not already been exploited are becoming rarer, which has led to what The New York Times has called a “emerging content crisis.”
This forced companies like Openai to conclude agreements with publishers to get even more gross material for their “hungry” chatbots. But, according to a forecast published last summer by the Epochai Research Institute, a lack of quality data for training LLM models could appear from 2026.
One of the most appreciated SF writers proposes Big Tech a measure worthy of his novels. And a warning: not to turn into “eloi”
An information problem on the Internet
As the chatting output is poured on the Internet, these “second generation” texts-which sometimes include manufactured information (called by “hallucinations” specialists) and flagrant errors-such as tips on pizza-will pollute the World Wide Web.
And if a chatbot “interacts” with the wrong type of content in the online environment, it can learn revolting ideas. Microsoft has found this for years before all the discussions about AI and Chatboţi. The American Tech giant decided in 2016 to stop “Tay”, a muzzle that began to reproduce racist and misogynist content.
Over time, all these issues could make the online information become less reliable and less useful than they are already. In addition, LLMs fed with “empty calories” content can produce even more problematic results, which in turn reach the web.
It is not difficult to imagine a vicious circle in which a continuous degradation occurs, as the chatbots feed on their own imperfect output.
A study published in Nature magazine in July 2024 explored the consequences of training AI on recursive data. The study showed that “irreversible defects” can lead to the “model collapse” for the systems thus trained. The easiest way to imagine this is with a copy of a photo, which is then copied, and so on, until any loyalty to the original image is lost.
How bad can the situation become?
Borges' story
Let's think of Borges's short story from 1941, the “Babel Library”. Fifty years before the computer scientist Tim Berners-lee created the architecture for the World Wide Web, Borges had already imagined an analog equivalent.
In this story of 36 pages, the writer outlines a world of a huge-and possibly infinite-of hexagonal rooms. On the shelves in each room are identical books as a structure – with the same number of pages, the same number of rows written on one page, all written with the same 25 -character alphabet.
These books contain, senses humanity, every possible permutation of letters for each spoken language.
Initially, this revelation causes joy: by definition, there must be books that detail the future of humanity and the meaning of life.
The inhabitants are looking for such books, only to discover that the vast majority contain only meaningless combinations. The truth exists – but there is every lie imaginable. And they are all buried in an amount of inconceivable of the gargle.
Even after centuries of searches, only a few significant fragments are discovered. And even then, there is no way to know if those coherent texts contain true or false information. Hope turns into despair.
A dilemma tied to the Internet
Will the Internet become so polluted that quality information will become a precious resource? Or will it produce an increasing number of chatbots so much vitiated content that finding the correct information in the online environment becomes as if you were looking for the AC proverbial in the hay car?
The characters in Stephenson's 2019 novel face this problem, subscribing to “edited flows”-news sources and information selected by people, considered trustworthy.
The disadvantage is that only the rich allow such personalized services, leaving the majority of humanity to consume online quality, unfiltered content.
“The Internet is often described as one of the great achievements of humanity. But, as any other resource, it is important to seriously reflect on how we maintain and manage it.
I hope you liked the special edition of the Nerd Alert section this weekend. In case you are curious to take a look over last Sunday, you can find it here:
The creator “Squid Game” launches the idea of a series “Spin-off” / Tom Cruise has established another world record with the new movie / “Resident Evil 9”
