Beyond the Steeper Curve: AI-Mediated Metacognitive Decoupling and the Limits of the Dunning-Kruger Metaphor

TL;DR

This paper challenges the simplistic view that AI amplifies the Dunning-Kruger effect, showing instead that AI can improve output but impair metacognitive accuracy, leading to a decoupling between performance and self-assessment.

Contribution

It introduces the concept of AI-mediated metacognitive decoupling, explaining how AI affects output quality and self-awareness differently, beyond the traditional Dunning-Kruger metaphor.

Findings

AI use improves short-term task performance.

AI degrades metacognitive accuracy and calibration.

AI widens the gap between output quality and self-assessed ability.

Abstract

The common claim that generative AI simply amplifies the Dunning-Kruger effect is too coarse to capture the available evidence. The clearest findings instead suggest that large language model (LLM) use can improve observable output and short-term task performance while degrading metacognitive accuracy and flattening the classic competence-confidence gradient across skill groups. This paper synthesizes evidence from human-AI interaction, learning research, and model evaluation, and proposes the working model of AI-mediated metacognitive decoupling: a widening gap among produced output, underlying understanding, calibration accuracy, and self-assessed ability. This four-variable account better explains overconfidence, over- and under-reliance, crutch effects, and weak transfer than the simpler metaphor of a uniformly steeper Dunning-Kruger curve. The paper concludes with implications for tool design, assessment, and knowledge work.