Artificial intelligence is capable of reducing mental effort, but the question arises regarding the consequences of frequently delegating cognitive processes to a machine. A study conducted at the Massachusetts Institute of Technology (MIT) found differences in brain activity, memory, and academic performance among young people who used ChatGPT to write texts.
Despite the significant resonance of the discovery, researchers emphasize that it does not allow for a conclusion about the inevitable emergence of 'cognitive debt.' According to the scientists, the answer lies not so much in the technology itself as in the way it is applied.
Delegating part of a mental task to an external tool is not a new phenomenon; calculators and search engines have long performed this function. Generative AI has only significantly expanded the spectrum of what can be transferred to a machine.
In an article published in The Conversation, Madaline Mariane Deliu Dumitru, a professor at the Faculty of Health Sciences at the Catholic University of Ávila and the University of Salamanca, together with Patricia Alzola Bordón, a neuropsychology graduate student from the University of Salamanca, links this behavior to the concept of cognitive delegation.
These researchers note: 'We use these tools so often that they almost become an extension of ourselves.'
How the study was conducted
Researchers from MIT observed 54 young people over three sessions. Participants were divided into three groups: one used ChatGPT, another used Google, and a third completed tasks without the help of external tools.
The group that used ChatGPT demonstrated lower scores in neural activity, linguistic aspects, and the quality of the generated texts. Furthermore, they showed poorer memory for written material and a weaker connection to their own ideas.
The experiment hypothesized that regular reliance on AI may reduce engagement in independent thinking.
The importance of motivation when using AI
However, the study did not examine an important aspect: why a person decides to assign a specific task to AI? A person might ask ChatGPT to write an essay because they feel incapable of doing so. But they might also use the tool to brainstorm ideas, search for alternative approaches, or improve an already started text.
Dumitru and Bordón argue: 'To understand the brain consequences of cognitive delegation, we must first ask why we delegate.'
This motive can significantly change the interpretation of technology use. Requesting a ready-made answer is not necessarily equivalent to using AI as a partner for studying a topic.
There is also another point: what happens to the time and energy that are no longer spent on completing the task? The authors refer to Google studies, which suggest that this tool can promote learning when used for information retrieval and knowledge expansion. A similar logic can apply to AI if it is used to discuss concepts, compare ideas, or support cognitive training.
Results of subsequent testing
The MIT experiment itself provided an important clue. In the fourth session, those who had previously used ChatGPT began working without it. Groups that used Google or relied solely on their own knowledge gained access to AI. The former continued to show reduced brain activity and poorer performance, while the others showed improvement in both metrics.
The so-called 'cognitive debt' remains just one possible hypothesis, not a final conclusion. Studies on the long-term impact of AI are still limited. Until more convincing evidence emerges, the way we use these tools seems as significant as the technology itself.