AI causes reduction in users’ brain activity – MIT
Key Takeaways
- Using LLMs like ChatGPT results in less cognitive engagement and reduced brain activity compared to unaided work or using search engines.
- Subjects using LLMs showed a dramatic decrease in 'ownership,' struggling to recall or summarize the content they produced.
- Long-term effects suggest that starting with AI leads to poorer performance in cognitive tasks when AI use is subsequently removed.
- Subjects who initially worked without technology and later introduced an LLM showed positive cognitive re-engagement when switching back to solo work.
- The researchers caution that the trend of replacing human thinking with LLMs poses a risk of diminishing long-term learning and critical thinking skills.
An MIT study involving EEG monitoring found that human brain activity significantly decreases when using LLMs for tasks like essay writing compared to unaided work or using search engines. The researchers observed lower neural connectivity and reduced cognitive load in AI users, alongside a dramatic drop in the 'ownership' or ability to recall and summarize their own output. When subjects switched roles later in the study, those transitioning from solo work to AI showed cognitive benefits, while those transitioning from AI to solo work exhibited weaker neural connectivity and under-engagement in key brain networks. The study concludes that relying on LLMs from the outset may lead to a long-term decline in essential learning and thinking skills, highlighting this as a pressing concern given the increasing use of AI in education and daily life.
