How AI Confidence Distorts Human Judgement

A new study from Aalto University reveals that using AI tools such as ChatGPT can make people perform better — yet less aware of how well they’ve really done. The familiar Dunning–Kruger effect reverses: AI-literate users become the most overconfident of all. This article explores how “cognitive offloading” and misplaced trust in machine reasoning can erode critical thinking and self-assessment, even among skilled professionals. Drawing on the research AI makes you smarter but none the wiser (Welsch et al., 2025), it examines the implications for training, workplace performance, and AI-enhanced learning, arguing for a shift from AI literacy to AI wisdom — where reflection and self-awareness matter as much as technical skill.

The Decline of Critical Thinking in the Digital Age

The Alarming Decline of Critical Thinking in the Digital Age Stay informed Subscribe to our newsletter. Newsletter Δ Subscribe Table of Contents Professor Elena Marquez stood before her university class, presenting a deliberately flawed study about caffeine consumption. She claimed the research showed definitive evidence that three cups of coffee daily would extend lifespan by […]