Ai boosts task scores but not users confidence calibration
Confident, Not Wiser: The Dunning-Kruger Effect in Human-AI Interaction
Human-Computer Interaction
Summary
People working with AI to solve reasoning problems perform better but don’t get better at judging how well they did. The study finds that both solo humans and humans using AI tend to overestimate their performance, and having AI help doesn’t improve this self-awareness. Confidence is less reliable at distinguishing right from wrong answers when AI is involved. The researchers suggest designing AI tools that help users check answers and understand AI’s accuracy instead of just providing final answers.
What this means in practice
- •For user interface designers: Create AI-assisted interfaces that better help users verify answers and understand AI reliability on specific tasks.
- •For knowledge worker teams: Support teams in evaluating the quality of combined human and AI work, not just accepting AI outputs as final.
Authors
Daniela Fernandes, Michelle Rausch, Agnes Mercedes Kloft, Daniel Buschek, Robin Welsch
Abstract
AI assistance can improve performance without improving self-assessment. We report a study (N=366) comparing Human alone and Human+AI performance on reasoning tasks, for which the AI model is benchmarked on the same items. Participants estimated global and block performance and rated confidence in their answers. Human+AI achieved higher scores, but self-estimates tracked performance weakly. Average overestimation was similar across groups, covering individual errors. Across tasks, confidence distinguished correct from incorrect answers less accurately in the Human+AI group, while within-task differences remained uncertain. The Dunning-Kruger pattern was found in both groups, with a larger observed contrast in Human+AI. Controls for score noise reduced but did not eliminate the pattern, with the controlled group difference remaining inconclusive. An extended computational account describes global and block estimates. Our findings distinguish performance augmentation from metacognitive augmentation and motivate interfaces that support verification, communicate task-specific AI model performance, and help users evaluate the quality of their joint work rather than produce answers.