MIT Study Examines How ChatGPT Affects Brain Activity During Writing

A study conducted by MIT looked at how much mental effort people use when writing with AI. The participants were divided into three groups. The first group used ChatGPT, the second group used a search engine, and the third group used no outside tool. Researchers used EEG to measure their brain activity.

The Brain-only group had the strongest neural connections, while the Search Engine group showed intermediate engagement. Those who used ChatGPT had the weakest brain connections. Researchers also found that the ChatGPT group had more difficulty remembering what they had written. They also reported feeling less connected to or responsible for their essays.

People who wrote on their own first and used AI later showed better memory and wider-ranging neural activity. The study suggests that doing your own thinking first and using AI later can lead to better mental engagement.

The study does not say that AI makes people stupid or causes brain damage. The study is still preliminary and had not been peer-reviewed at the time mentioned in the document. It also had a small number of participants, and the results may not apply to every person, AI model or type of task.

The three groups were the LLM group, Search Engine group and Brain-only group. A total of 54 participants took part in the first three sessions, while 18 participants completed the fourth session. EEG was used to record the participants’ brain activity, while NLP analysis was also carried out.

Many participants were interviewed after each session. Each participant was assessed through an essay-writing task, which was evaluated with the help of human teachers and an AI judge.

Those in the Brain-only group showed the strongest and widest-ranging neural networks. Those who used a search engine showed intermediate engagement, while the LLM group showed the weakest overall coupling.

The LLM group performed worse than the Brain-only group at the neural, linguistic and scoring levels. The researchers describe these findings as preliminary.

The paper had not yet been peer-reviewed as of June 2025. The study also had several limitations, including a limited number of participants.

The study’s results cannot be directly applied to other LLMs. The study focused only on essay writing in an educational setting, so the results may not apply to other tasks. Researchers now want to study the long-term effects of AI use on memory, creativity and writing skills.

Importantly, the researchers did not conclude that AI makes people “dumb” or causes “brain damage”. They specifically advised against using such terms when describing the study.