AI Devours the Web: The Internet's Collective Memory Is Fading
Original: As AI eats the web, the internet’s collective memory is disappearing
Why This Matters
Search infrastructure decay threatens the reliability of AI systems trained on and summarizing an increasingly corrupted web.
Google's AI-powered search summaries are generating factual errors—such as fabricating sunset times—while link rot, coding errors, and AI-generated content pollution are collectively eroding the web's reliability as a shared repository of accurate information.
A growing body of anecdotal evidence and expert observation suggests Google Search is faltering as the web's primary truth infrastructure. Users in Colorado Springs and elsewhere have reported Google's AI Overviews inventing incorrect sunset times, causing real-world confusion. Tech observers note Google has 'lost its edge' returning accurate results. The article, written by Vass Bednar for The Walrus, argues the problem goes beyond AI hallucinations or 'enshittification.' Structural decay is accelerating: link rot removes pages daily, the Library of Congress briefly lost sections of the U.S. Constitution to a coding error, and 404 Media has reported companies planting artificial content on Reddit specifically to manipulate AI-generated search summaries. Google's AI layer, by interposing between users and primary sources, can render existing pages practically undiscoverable. Bednar frames this as a question of cultural sovereignty—arguing that the dominant policy debate around foreign platforms like Netflix or YouTube obscures the deeper issue of who preserves and controls the digital public record. The article calls for broader recognition that the web's collective memory is not self-sustaining and is deteriorating in real time.