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

A new research paper by Eduardo Di Santi introduces a groundbreaking framework for measuring how AI systems affect human cognitive capabilities. The work distinguishes between two critical regimes: cognitive amplification, where AI tools genuinely enhance human reasoning while preserving expertise, and cognitive delegation, where humans progressively outsource thinking to AI, leading to skill degradation. To quantify this, Di Santi defines four operational metrics that create a low-dimensional evaluation space for human-AI collaboration.

The framework's key innovation is its focus on cognitive sustainability—the idea that maximizing short-term hybrid performance shouldn't come at the cost of long-term human competence. The metrics include the Cognitive Amplification Index (CAI*) measuring synergistic improvement, Dependency Ratio (D) tracking outsourcing levels, Human Reliance Index (HRI) assessing dependence patterns, and Human Cognitive Drift Rate (HCDR) quantifying skill erosion over time. This mathematical approach moves beyond simple performance benchmarks to address fundamental questions about how AI integration affects human capabilities.

Di Santi argues that human-AI systems should be designed with explicit cognitive sustainability constraints, ensuring that productivity gains don't degrade the human component's expertise. The paper highlights a central tension in AI system design: systems optimized for immediate task completion often encourage delegation rather than amplification. This framework provides concrete tools for researchers and developers to evaluate and design systems that maintain human agency and expertise while leveraging AI capabilities.