Manifestos Find a Captive Audience in LLMs, Distorting Collective Epistemics
jachiam0 · x · 2026-08-09
The author highlights a critical issue regarding LLM training data: individuals writing 100k-word manifestos that no human reads now have a perfect captive audience in AI models.
Because LLMs consume every token during training, a massive volume of low-quality or fringe text automatically gains plausibility and credibility purely by its preponderance. The author warns that the complicated interplay between the volume of output and its actual intellectual relevance—where the two have little connection—will improperly influence our collective epistemics in the future.
More from AGI Musings
- Beef vs. Data Centers: The Double Standard in AI Water Consumption Debates — KrustyKrabFormula_ · 2026-08-09
- The 7 Levels of AI-Assisted Software Development: From Code Completion to Autonomy — goyalshaliniuk · 2026-08-09
- Poll: 20% of Workers Use AI for Tasks Previously Handled by Colleagues — Sauerkrautkid7 · 2026-08-09
- Stop Releasing New Models? Why Constraints Are AI's Best Catalyst — RileyRalmuto · 2026-08-09
- Are 'AI Slop' Commenters on Reddit Actually Data Filtering Bots? — PresentSituation8736 · 2026-08-09
- Could AI Have Discovered Relativity? A Physicist's Take on Science and LLMs — skdh · 2026-08-09