Jorge Luis Borges' 1941 short story "The Library of Babel" may have predicted a disturbing fate for the modern internet - one where finding reliable information becomes nearly impossible amidst an ocean of AI-generated content.
In his prescient tale written decades before the web's invention, Borges imagined a vast library containing every possible combination of letters. While this initially excited its inhabitants who believed they could find books containing all truths, they soon discovered that meaningful content was buried in an effectively infinite amount of gibberish.
This fictional scenario bears an unsettling resemblance to current concerns about AI chatbots and their impact on online information. As these language models consume existing high-quality web content for training, they produce new text that often contains errors, fabrications, and "hallucinations" - which then pollutes the internet further when published online.
Experts warn this could create a dangerous feedback loop. When chatbots are trained on increasingly low-quality content, including their own flawed outputs, the resulting degradation could make finding accurate information as challenging as locating specific books in Borges' infinite library.
A recent Nature study suggests this recursive training process leads to "irreversible defects" and "model collapse" in AI systems - similar to how repeatedly copying a photo results in loss of quality.
Some fear this could create a two-tiered internet where only wealthy users can access reliable, human-curated information behind paywalls, while most people are left to navigate an increasingly polluted sea of AI-generated content.
Science fiction author Neal Stephenson explored this scenario in his 2019 novel "Fall," depicting a future where most of the internet becomes unusable due to misinformation, with only the rich able to afford trustworthy "edit streams."
As AI companies face a predicted shortage of high-quality training data as soon as 2026, Borges' cautionary tale serves as a reminder that the internet's value lies not in its size, but in our ability to find truth within it.
Note: The provided link about swatting is not contextually relevant to the article's topic of AI, information quality, and Borges' story, so following instruction #4, I have omitted it.