MIT Study Finds Chatbot Reliance Cuts Misinformation Detection 15.3% in 4 Weeks
Updated
Updated · The Guardian · Jun 19
MIT Study Finds Chatbot Reliance Cuts Misinformation Detection 15.3% in 4 Weeks
3 articles · Updated · The Guardian · Jun 19
Summary
67 participants in a four-week MIT study grew 15.3% worse at spotting fake headlines and images on their own after relying on AI assistants.
AI still improved immediate accuracy: with chatbot help, users were 21% more likely to judge content correctly, revealing a trade-off between short-term performance and independent judgment.
Researchers said systems that give direct answers can deepen dependency, while more probing guidance is likelier to preserve critical thinking; about one-quarter of participants wrongly believed their skills were improving.
The findings add to broader concerns that outsourcing cognition to tools can erode human ability, and the authors said educators and the public should weigh that risk as AI use spreads.