CONCEPT
The Memetic Feedback Loop
The unprecedented cycle by which AI systems trained on the human meme pool generate outputs that re-enter the meme pool, creating a feedback dynamic with no precedent in the 70,000-year history of cultural evolution.
When a large language model is trained on the textual output of human civilization, it is, in
Richard Dawkins’s terms, absorbing a snapshot of the
meme pool—the total body of cultural information that has survived the selection pressures of human attention, editorial judgment, and institutional curation to exist in written form. The patterns the model extracts are memetic patterns: regularities in how ideas are expressed, connected, and deployed across the vast landscape of human discourse. When the model generates output and a user accepts it, builds on it, incorporates it into a document or a codebase or a published argument, that output re-enters the meme pool—where it influences future human thinking, is potentially absorbed into future AI training data, and may influence future AI outputs in turn. This is the memetic feedback loop: a cycle of generation, selection, and reinjection that has no precedent in the 70,000-year history of cultural evolution, because no previous substrate for the meme