When machines write most of the internet: what happens to human knowledge
As AI-generated text floods the web and future models train on it, examine the risk of a feedback loop that degrades quality — and what it takes to keep genuine human knowledge findable.
Listen in the Fylom app.
Thirty-five percent of new websites are now generated or assisted by artificial intelligence.
AI content is one hundred and seven percent more likely to exhibit positive sentiment than human writing.
Recursive training on synthetic data causes model collapse, turning nuanced information into repetitive gibberish.
Answering engines reduce traffic to original publishers by forty percent, threatening the survival of investigative journalism.
The liar's dividend allows bad actors to dismiss real evidence as deepfakes to escape accountability.
Future internet structures may bifurcate into a premium human layer and a synthetic commons.
- 01Intro1 min
- 02The Great Dilution: Measuring the AI Influx2 min
- 03Model Collapse and the Ouroboros Effect3 min
- 04The Death of Discovery Signals3 min
- 05Preserving the Human Provenance2 min
- 06Outro1 min
- The Impact of AI-Generated Text on the Internet
- AI search could kill the web without new quality signals and revenue models
- Model collapse and synthetic data: what happens when the internet fills with generated content
- Model Collapse Is Coming for AI Search — And Nobody’s Talking About It | Towards AI
- AI Slop and the End of Knowledge | Psychology Today
- AI content flood: why the web's signal is dying | Psyll
- AI Search Is Erasing the Signals That Made Quality Content Discoverable | FrontierNews.ai
- What's Next for the AI Ouroboros? | The AI Journal
- than half of new articles on the internet are being written by AI
- AI Search Cuts Publisher Clicks 40% — and Erodes Discovery Signals
- The Impact of AI-Generated Text on the Internet
- The Impact of AI-Generated Text on the Internet
- The Impact of AI-Generated Text on the Internet
- Researchers find AI text is making the internet more uniform and weirdly cheerful
- Dead Internet? A Third of New Websites Are AI-Generated, Says Stanford - Decrypt
- The Impact of AI-Generated Text on the Internet
- AI models collapse when trained on recursively generated data | Nature
- Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating World
- LLM Web Dynamics: Tracing Model Collapse in a Network of LLMs
- Collapse or Thrive?Perils and Promises of Synthetic Data in a Self-Generating World
- Retrieval Collapses When AI Pollutes the Web | Proceedings of the ACM Web Conference 2026
- Retrieval Collapses When AI Pollutes the Web
- The Internet is Eating Itself, and AI is Holding the Fork | Harvard Business School AI Institute
- Collapse or Thrive?Perils and Promises of Synthetic Data in a Self-Generating World
Fylom generates episodes like this on any topic you're curious about.
Fylom episodes are researched, written, and voiced by AI. Automated checks help catch inaccuracies, but episodes aren't reviewed by a human and AI can still get things wrong. Treat them as a starting point, not a source of record — more in our accuracy disclaimer.