Recent research indicates that the brains of individuals who utilize ChatGPT for writing essays are less active compared to those who write without any digital aids. This study is part of a growing effort to determine if artificial intelligence (AI) is contributing to cognitive laziness.

At the MIT Media Lab in Cambridge, Massachusetts, computer scientist Nataliya Kosmyna and her team tracked brain activity in college students tasked with essay writing. Participants used either a chatbot, a traditional internet search, or no internet tools at all. While the primary findings were expected, the research also suggested that initial reliance on a chatbot could lead to decreased brain engagement, which persists even after the tool is no longer used.

Kosmyna emphasized the importance of not overinterpreting these preliminary results. She humorously remarked that the study doesn’t imply any “dumbness in the brain, no stupidity, no brain on vacation.” The study involved a limited number of participants over a brief period and does not conclusively speak to long-term cognitive effects from regular chatbot usage, nor does it address how the brain might behave in other AI-assisted scenarios. “We don’t have any of these answers in this paper,” Kosmyna stated. The findings were released on the preprint platform arXiv on June 10, before peer review.

On supporting science journalism

If you find this article engaging, consider supporting our award-winning journalism by subscribing. Your subscription helps ensure the continuation of influential stories that explore the discoveries and ideas shaping our world today.

Easy essays

Kosmyna’s group enlisted 60 students aged 18 to 39 from five universities in the Boston area. The participants were asked to spend 20 minutes writing short essays on topics commonly found on the Scholastic Assessment Tests (SATs), such as “should we always think before we speak?”

The students were divided into three groups: one used ChatGPT, powered by OpenAI’s GPT-4o model, as their only information source; another group used Google for research without AI assistance; and the last group was prohibited from using the Internet. After writing essays on three questions within their assigned groups, 18 participants were switched to a different group to compose a fourth essay on a previously discussed topic.

Each participant was equipped with a commercial EEG cap, which recorded brainwave activity during the essay writing process. These caps track minute voltage fluctuations that indicate which regions of the brain are interacting.

Notably, the students who wrote essays without internet assistance displayed the most robust and extensive brain connectivity, particularly from the back to the front of the brain, which is associated with decision-making. They were also better at recalling details from their essays during follow-up questions.

The group using Google showed heightened activity in areas linked to visual processing and memory. Meanwhile, the ChatGPT users had the lowest levels of brain connectivity during the task.

Kosmyna noted that increased brain connectivity isn’t necessarily beneficial or detrimental—it could indicate deeper engagement with a task, cognitive inefficiency, or an overwhelming cognitive load.

Creativity lost?

Intriguingly, those who initially used ChatGPT and later wrote without digital tools showed increased brain connectivity, though not to the same extent as those who had foregone digital tools entirely from the start.

“This finding confirms some concerns among creativity researchers regarding the potential negative impact of AI on fundamental creative processes,” mentioned Adam Green, co-founder of the Society for the Neuroscience of Creativity and a cognitive neuroscientist at Georgetown University in Washington DC. However, Green highlighted that only 18 participants were involved in this part of the study, introducing some uncertainty to these observations.

Interestingly, the study also found that using a chatbot to rewrite an essay previously done without digital aids actually increased brain connectivity, contrary to expectations. This suggests that the timing of introducing AI tools in learning contexts might be critical, as per Kosmyna.

Many experts in education are hopeful about the potential of chatbots as personalized educational tools. Guido Makransky, an educational psychologist at the University of Copenhagen, mentioned that these tools are most effective when they encourage students to reflect rather than simply providing answers.

This article is reproduced with permission and was first published on June 25, 2025.