The Synthetic Loop: How AI is Starving the Open Web
AI-generated image · Bay Street Wire
As Google replaces sources with summaries and corporations pollute the public record, the internet's archival function is collapsing into a feedback loop of 'slop.'
From a practitioner's perspective, the current trajectory of search isn't just a matter of 'sloppy' models or a few hallucinations; it is a systemic failure of the internet's knowledge infrastructure. We are witnessing the transition from a durable, human-curated web to a degraded digital space where the original sources are becoming undiscoverable.
As The Walrus reports, Google's integration of AI summaries is fundamentally altering the relationship between the user and the source. By inserting an error-prone AI layer between the query and the original page, Google has created a environment where factual information—even something as basic as sunset times in Colorado Springs—is invented, rendering existing underlying pages practically invisible.
This is not merely a technical glitch; it is a collapse of the corpus. The Walrus notes that the very infrastructure designed to preserve our collective memory is breaking down. The Internet Archive, which operates the Wayback Machine as a critical fail-safe for the web, is currently under immense pressure from cyberattacks and expensive litigation. Furthermore, news organizations are now blocking Wayback Machine crawlers to prevent AI companies from using archived pages as indirect sources of copyrighted material.
Even the most resilient pillars of open knowledge are feeling the strain. The Walrus reports that Wikipedia is facing a crisis because AI systems now scrape and ingest its content directly. By presenting results without requiring users to click through to the site, AI has turned Wikipedia into the infrastructure of its own demise, cutting off the traffic and donations necessary for its survival.
Worst of all, the remaining human-curated spaces are being intentionally contaminated. 404 Media reports that companies are planting content on Reddit specifically to manipulate the answers generated by AI search. This creates a dangerous feedback loop: AI models draw summaries from a public record that is being poisoned by synthetic 'slop,' which in turn informs the next generation of AI outputs.
When combined with the corporate erasure of archives—such as The Walt Disney Company deleting the FiveThirtyEight archive after laying off its staff in March 2025—the result is a digital landscape where truth is not just harder to find, but is actively being erased. We are trading a stable body of knowledge for a system that prioritizes efficiency over accuracy, risking a future where the models we build have no authentic human data left to consume.