MIT Study: Heavy AI Reliance Weakens Users' Ability to Detect Misinformation

A four-week MIT Media Lab study with 67 participants found that while AI tools were effective at identifying fake news headlines and images, users who relied heavily on AI answers became worse at spotting misinformation independently over time. Researchers attribute this to AI delivering confident verdicts rather than guiding users through the reasoning process, meaning the underlying critical-thinking skill went unpracticed. The effect mirrors how GPS navigation has gradually diminished people's natural sense of direction. The findings, published in the CHI 2026 conference proceedings, are consistent with a broader body of recent research showing that AI reduces cognitive effort on tasks but can simultaneously erode the reasoning skills that effort would have developed. Experts suggest using AI as an active thinking partner — by questioning its reasoning or asking what could make its answer wrong — rather than passively accepting its conclusions.