You are currently viewing The neural impact of Artificial Intelligence

Does AI really help us, or does it simply let us skip the exhausting work of critical thinking by doing the thinking for us?

A recent study from MIT’s Media Lab set out to quantify exactly how AI affects our brains. The researchers divided 54 participants, aged 18 to 39, into three groups and asked each to write an essay.

The first group had only OpenAI’s GPT at their disposal, no other resources, no web browsers. The second group could use any website to aid their writing, but all large language models (LLM), including ChatGPT, were strictly off-limits. The third group had no help at all: no browser, no AI, just their own minds.

 

As each participant wrote, a 32-channel electroencephalography (EEG) headset recorded brain activity across different regions, allowing the scientists to analyze how each approach affected the brain in real time.

 

The results were striking. EEG data showed clear differences in brain connectivity between the groups. Those who relied solely on their own abilities activated a broad network of brain regions. The browser group’s results were somewhat diminished, but still robust. The ChatGPT group, however, showed a dramatic 55% reduction in brain activity in areas linked to memory, problem-solving, and attention. 

 

In other words, participants using ChatGPT had the lowest brain engagement and, as the researchers put it, “consistently underperformed at neural, linguistic, and behavioral levels” compared to the brain-only group, which achieved the highest scores across the board.

To further test these findings, participants were asked to quote passages from their own essays. Both the brain-only and browser groups had no trouble recalling what they had written. But among those who used ChatGPT, 83% of participants failed to prove a correct quotation. 

 

The study also tracked how brain connectivity evolved over four months of essay writing. The group working without external aids demonstrated rapid cognitive adaptation, with growing activation of memory, organizational, and control networks through practice. In contrast, the group using AI exhibited lower brain activity over time, especially in regions responsible for planning and memory. This suggests that relying on AI offloads cognitive effort, reducing the brain’s involvement in organizing and generating ideas.

 

Scientists have defined a term for this phenomenon : cognitive offloading. It describes the tendency of individuals to depend on AI tools for parts of a cognitive task, leading to a measurable decline in brain connectivity across key areas.

 

The bottom line : frequent use of AI may hinder the development of brain networks essential for critical thinking and independent work, potentially resulting in writing that is less original and less deeply reasoned once the AI support is removed. The researchers recommend a balanced approach: use AI for routine assistance, but make sure to practice writing and thinking independently to maintain and strengthen your own cognitive abilities.

 

Liza Samaha 

Cyberjustice, M2 

 

https://time.com/7295195/ai-chatgpt-google-learning-school/?utm_source=chatgpt.com

https://arxiv.org/pdf/2506.08872v1

https://www.media.mit.edu/publications/your-brain-on-chatgpt/



Cette publication a un commentaire

  1. Samaha

    I agree on the article, to my opinion, the use of the AI for routine assistance, is a great help in the day today tasks, but we have to make sure to practice writing and thinking independently to maintain and strengthen our own cognitive abilities.

Laisser un commentaire

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur la façon dont les données de vos commentaires sont traitées.