This article was automatically translated from the original Turkish version.
We all occasionally turn to AI tools like ChatGPT to quickly gather information on a topic, write an article, or solve a complex problem. These tools make life easier and save us time. But what about the hidden costs? What if this convenience is weakening our brain’s ability to learn and remember? The latest study by the MIT Media Lab sheds light on a closely relevant truth by seeking answers to these questions: learning with AI may be more costly than we think…
The preprint article titled “Your Brain on ChatGPT,” conducted at the MIT Media Lab, examines how people and organizations interact daily with LLM products like ChatGPT and how these interactions affect our cognitive processes.
Participants in the study were divided into three groups: the LLM group, the Search Engine group, and the Brain-only group. Each group completed a writing task using their assigned tool (or no tool at all). Researchers recorded participants’ brain activity using electroencephalography (EEG), a method used to assess cognitive engagement and cognitive load, and to gain deeper insight into neural activation patterns during the writing task. The results showed that each group exhibited significantly different neural connectivity patterns, reflecting distinct cognitive strategies.
One of the most striking findings was that neural connectivity systematically decreased in proportion to the amount of external support used. The “Brain-only” group, which relied solely on their own minds, displayed the strongest and most comprehensive networks. The Search Engine group showed moderate levels of connectivity. The weakest overall connectivity was observed in the group using LLM assistance.
After four sessions spanning four months, participants in the LLM group performed worse than their “Brain-only” counterparts at the neural, linguistic, and scoring levels. Additionally, the LLM group reported a weaker sense of ownership over the articles they wrote and struggled to quote from their own texts just minutes after writing them.
Neural Representation of LLM, Search, and Brain Groups in the Article from the Study “Your Brain on ChatGPT” (Mit)
While this pioneering study presents the first findings on the potential impact of AI on learning skills, it also has limitations. The study involved a small number of participants and used only ChatGPT. Therefore, the results cannot be generalized to other LLM models. Moreover, since this article has not yet undergone peer review, the findings must be interpreted with caution.
So what do these findings tell us? While these tools offer practical benefits in the short term, they may lead to a decline in our learning abilities over the long term. As AI’s role in our lives continues to expand, we must pay attention not only to its practical advantages but also to its potential effects on our cognitive processes. The “Your Brain on ChatGPT” study provides a scientific foundation for this debate by highlighting the cognitive cost of using AI-assisted tools.
Building on this research, here are some additional reflections on our future relationship with AI.
Learning and Reasoning Skills: The convenience offered by LLMs may distance us from the need to solve problems and think deeply. The study shows that the LLM group had weaker neural connections compared to both the Search Engine and Brain-only groups. This suggests that constant reliance on external tools may weaken our cognitive strategies and neural networks. Just as a muscle weakens when unused, our mental abilities can become passive through overreliance on such tools.
Creativity and Originality: LLMs can accelerate text generation and idea production. However, the study’s findings reveal that LLM users feel less ownership over the texts they produce. This raises important questions about how creativity and originality are affected by LLM use. How can these tools help us develop our own original ideas rather than simply generating content for us?
The Future of Education: As the use of AI in education becomes inevitable, understanding its potential risks is critical. The study’s authors note that the impact of LLMs on education is “only just beginning to take shape.” Therefore, educators and students must develop new strategies to use these tools consciously. How can AI be integrated not merely as a source of information but as a tool that enhances critical thinking and problem-solving skills?
To improve the thinking process, it may be helpful to consciously manage your cognitive strategies and experiment with new approaches. Based on the findings of the “Your Brain on ChatGPT” article, here are some methods to enhance your thinking process:
Reduce dependence on external support: The study shows that the group using LLM assistance had the weakest overall neural connectivity. This suggests that constant reliance on external tools may reduce cognitive engagement. Therefore, when faced with a complex task, focus first on using your own mental resources before turning to an AI tool.
Adopt the “Brain-only” approach: In the study, the group labeled “Brain-only” exhibited the strongest and most comprehensive neural networks. This indicates that relying initially on your own knowledge and thoughts when researching a topic, writing an article, or solving a problem is beneficial.
Separate subtasks like idea generation and writing: One limitation of the study was that the writing task did not distinguish between subtasks such as idea generation and writing. To improve your thinking process, break tasks into smaller components. For example, when writing an article, first generate ideas and outline the structure on your own before moving to the writing phase. This can help ensure deeper cognitive engagement at each stage.
Strengthen ownership of knowledge: The study found that participants in the LLM group felt less ownership over their written articles and struggled to quote from them. Instead of passively receiving information, rephrase it in your own words and internalize it. This makes the knowledge more durable in your memory.
Use different tools strategically: Search engines prompted greater cognitive engagement than LLMs. Therefore, when researching a topic, it may be more beneficial to use search engines to gather information from diverse sources and synthesize it yourself, rather than relying on AI for ready-made answers. This encourages your brain to work more actively and enhances your learning skills.
Continuous learning and practice: The article raises concerns that LLM use may lead to a long-term decline in learning abilities. To improve your thinking process, engage in activities that keep your mind active. Reading, learning new skills, solving puzzles, or undertaking creative projects will strengthen your cognitive abilities.