AIBearisharXiv – CS AI · 6h ago7/10
🧠
Human-AI Co-Evolution and Epistemic Collapse: A Dynamical Systems Perspective
Researchers propose a unified dynamical systems model of human-AI co-evolution, showing that increased reliance on LLMs creates feedback loops between human cognition, data quality, and model capability. The analysis identifies three regimes including a 'degenerative convergence' where over-reliance on AI leads to reduced diversity and an information bottleneck, suggesting AI trajectory depends as much on human behavioral dynamics as on model design.