Does Using ChatGPT Change Your Brain Activity? Study Sparks Debate
Scientists caution against overinterpreting a small experiment regarding ChatGPT and brain activity that has garnered significant attention.
Thai Liang Lim/Getty Images
Research indicates that the brain activity of individuals using ChatGPT for essay writing is less engaged compared to those restricted from using any online resources. This study is part of a larger investigation into whether AI contributes to cognitive laziness.
Nataliya Kosmyna, a computer scientist at the MIT Media Lab in Cambridge, Massachusetts, along with her team, examined brain-wave activity in university students as they composed essays using either a chatbot, an Internet search tool, or no online assistance. While the primary findings were expected, some observations were notably intriguing, such as hints that dependence on a chatbot for initial drafting could result in lower brain engagement, even when the tool is removed later.
Kosmyna emphasizes caution against over-interpretation of the results. The study does not conclusively indicate “dumbness in the brain; no stupidity, no brain on vacation,” she quips. With only a few dozen participants over a short duration, it cannot determine if regular chatbot use alters long-term cognitive functions or how the brain responds during varying AI-assisted tasks. “We don’t have any of these answers in this paper,” Kosmyna states. The findings were shared on the preprint server arXiv on June 10.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. Your subscription helps ensure the continuation of impactful stories regarding the discoveries and ideas that are shaping our world today.
Easy essays
Kosmyna’s research team gathered 60 students, aged 18 to 39, from five Boston-area universities. The participants spent 20 minutes crafting short essays responding to questions resembling those found on Scholastic Assessment Tests (SATs), such as “Should we always think before we speak?”
The students were divided into three groups: one used ChatGPT as the sole information source for their essays; another utilized Google for research without AI assistance; and the last group was entirely offline. Ultimately, 54 participants wrote essays in their respective groups before 18 were reassigned to a new group to tackle a fourth essay based on a previously discussed topic.
Each participant donned a commercial electrode-covered cap that collected electroencephalography (EEG) data during the writing process. These devices measure slight voltage fluctuations from brain activity and reveal which brain areas are ‘communicating’ with one another.
Students crafting essays using solely their cognitive resources demonstrated the strongest connectivity among brain regions, with heightened activity flowing from the rear to the front regions of the brain responsible for decision-making. Additionally, they were, as anticipated, better at recalling their own content when questioned by researchers.
Conversely, the Google group exhibited increased activity in regions linked to visual processing and memory, while the chatbot group showed the lowest brain connectivity during the task.
Increased brain connectivity isn’t necessarily indicative of better or worse engagement, Kosmyna notes. Generally, higher brain activity could suggest deeper task engagement, inefficiency in thought processes, or even cognitive overload.
Creativity lost?
Interestingly, participants who initially utilized ChatGPT and later switched to writing without any online tools exhibited increased brain connectivity; however, this did not reach the levels seen in participants who wrote without tools from the start.
“This evidence aligns with concerns voiced by various creativity researchers about AI — excessive reliance on AI for idea generation may hinder practice in fundamental creative processes,” notes Adam Green, co-founder of the Society for the Neuroscience of Creativity and a cognitive neuroscientist at Georgetown University.
Green points out that the inclusion of only 18 participants in this aspect of the study introduces uncertainty into the findings. Other factors could have influenced the results, such as students rewriting essays on previously discussed topics, possibly tapping into different cognitive resources than those required for entirely new subjects.
Adding to the complexity, the study revealed that transitioning to a chatbot for essay writing after having previously done so without any online tools actually enhanced brain connectivity — contrary to expectations. This indicates the importance of considering when AI tools should be introduced to learners to improve their experience, Kosmyna says. “The timing may be significant.”
Many educational experts express optimism regarding chatbots as effective, personalized tutors. Guido Makransky, an educational psychologist at the University of Copenhagen, asserts that these tools are most beneficial when they encourage students to ask reflective questions rather than simply providing answers.
“It’s an interesting paper, and I understand the attention it’s receiving,” Makransky remarks. “However, in a practical context, students would and should engage with AI differently.”
This article is reproduced with permission and was first published on June 25, 2025.
