New Metric Framework Distinguishes AI Amplification from Delegation

A recent study published on arXiv (2603.18677) presents a theoretical and mathematical framework aimed at distinguishing cognitive amplification from cognitive delegation in human-AI interactions. Cognitive amplification refers to the enhancement of performance in hybrid human-AI systems while maintaining human expertise, whereas cognitive delegation denotes the gradual transfer of reasoning tasks to AI, potentially leading to a decline in human skills over time. The framework outlines four key metrics: the Cognitive Amplification Index (CAI*), assessing collaborative improvement beyond the most effective individual agent; the Dependency Ratio (D) and Human Reliance Index (HRI), which evaluate AI's structural influence on hybrid outputs; and the Human Cognitive Drift Rate (HCDR), indicating changes in human performance over time. This study lays the groundwork for identifying beneficial AI integration versus detrimental dependency.