Research: Long-term reliance on large language models like ChatGPT may impair cognitive abilities

ChainCatcher news, a research team from MIT, Wellesley College, and the Massachusetts College of Art and Design has recently discovered that long-term reliance on large language models (LLMs) like ChatGPT may impair cognitive abilities.

The study shows that participants using AI for writing have only half the number of Alpha wave connections in their brains compared to independent thinkers (42 vs. 79), and 83.3% are unable to accurately cite their own article content. The researchers refer to this phenomenon as "cognitive debt"—which alleviates the mental burden in the short term but leads to a decline in critical thinking and creativity in the long term.