Cognitive Amplification vs Cognitive Delegation in Human-AI Systems: A Metric Framework

A new framework has been developed to differentiate between cognitive amplification and cognitive delegation within human-AI systems. This framework introduces four metrics—the Cognitive Amplification Index (CAI*), Dependency Ratio (D), Human Reliance Index (HRI), and Human Cognitive Drift Rate (HCDR)—to evaluate whether AI integration enhances human cognitive capabilities or fosters excessive reliance. The framework's utility was tested using an agent-based simulation.

Why it matters

The distinction between AI as a tool for cognitive amplification versus cognitive delegation is critical for designing effective and sustainable human-AI systems. Understanding these dynamics is essential for preserving and enhancing human expertise, preventing over-reliance, and optimizing overall system performance in critical decision-making contexts.

Key insights

• Human-AI system design often blurs the line between AI enhancing human cognition and AI leading to human over-reliance or degradation of capability.
• A novel framework is proposed to analytically distinguish cognitive amplification from cognitive delegation.
• The framework defines four quantitative metrics: Cognitive Amplification Index (CAI*), Dependency Ratio (D), Human Reliance Index (HRI), and Human Cognitive Drift Rate (HCDR).
• These metrics aim to assess system designs that improve hybrid performance without diminishing human capability versus those that encourage outsourcing of reasoning to AI.
• The framework was empirically tested through an agent-based NetLogo simulation under various reliance and dependency-atrophy configurations.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2603.18677