AI makes you 15% worse at spotting fake news
mit media lab researchers found participants became 15% worse at identifying fake news after four weeks of relying on chatbots, revealing a critical cognitive dependency that undermines media literacy.
researchers at the mit media lab tracked users detecting fake news over four weeks, finding that while ai assistance initially boosted accuracy by 21%, it eventually caused users to become 15% less accurate without it than before. this 'ai dependency paradox' indicates users forget that llms are statistical models rather than infallible oracles, creating over-confidence. however, the study found that socratic ai interfaces that 'ask' rather than 'tell' can mitigate this erosion of critical thinking. this signals a need for product designers to prioritize human retention over speed, favoring interaction styles that train users rather than replacing their skills.