Writing negatively affects brain activity
A recent study by the Massachusetts Institute of Technology (MIT) revealed that relying on generative artificial intelligence tools such as ChatGPT for writing tasks leads to a noticeable decline in brain activity and cognitive functions, compared to those who use search engines or rely on their own writing skills. Study details
Researchers from the MIT Media Lab conducted a four-month study in which they compared three groups of participants while completing written tasks. The results showed that people who used ChatGPT completed their writing tasks faster than others by 60%, but this came at the expense of the mental effort associated with understanding and organizing information in long-term memory. They recorded a 32% decrease in the so-called German Cognitive Load, an indicator of how deeply the brain absorbs information and organizes it in long-term memory. The study also showed that the articles written by the group that used ChatGPT were remarkably similar and lacked originality, and the participants expressed a weak sense of belonging or “ownership” towards what they wrote. With repeated use of the tool, the researchers noticed a gradual deterioration in performance, users became content with copying the generated texts without revision or critical thinking. These negative effects persisted even after stopping using the tool, suggesting the possibility of permanent changes in the way the brain works. The brains of young people, who are still developing, are expected to be more susceptible to these effects, raising concerns about the widespread use of artificial intelligence tools in educational environments. Changes in neural activities in the brain
The study relied on electroencephalographic (EEG) scans to monitor neural activities. The results showed that participants who wrote based on their own abilities had more interconnected neural connections than others; they recorded 79 neural connections in the range of alpha waves associated with concentration and creative thinking. Those who used search engines achieved an average level of performance, and ChatGPT users recorded the weakest performance with only 42 connections. A similar decrease was also observed in the Theta (theta) wave range, associated with memory and executive control; the number of neural connections in the group that relied on their own writing skills was 65, compared to only 29 neural connections in ChatGPT users. These differences suggest that there is an inverse relationship between relying on artificial intelligence tools and the brain's involvement in information processing. Memory dysfunction and decreased ability to remember One of the most alarming findings is that 83% of ChatGPT users could not remember quotes from articles they wrote Just minutes ago, and the percentage was only 11.1% among those who used search engines or wrote without assistance. When asked to rewrite the article without using artificial intelligence, they were unable to remember most of the content, indicating poor information processing in long-term memory. Future effects in education
The results of this study raise fundamental questions about the danger associated with the widespread use of artificial intelligence tools in educational environments, especially among young age groups that are still developing their mental abilities. The lead researcher of the study, Natalia Kosmyna, warned that students who rely on tools such as ChatGPT may develop different cognitive patterns that affect their future mental skills. These findings are consistent with other studies that suggest that intensive use of artificial intelligence may contribute to increased loneliness and decreased creativity, even as it contributes to improved productivity.