What Artificial Intelligence Is Doing to Our Brains
The brain adapts to the tools we use, just as it once adapted to writing and the calculator.
With AI, this adaptation is faster and deeper, because the tool does not perform just a single task, it reaches into the very core of thinking: reasoning, memory, and decision-making.
A key scientific finding: Researchers at the MIT Media Lab (Kosmyna et al., 2025) tracked 54 participants over four months using EEG devices while they wrote essays, in a study titled "Your Brain on ChatGPT." The group that wrote entirely on their own showed the strongest connectivity between brain regions responsible for memory and reasoning. The group that used ChatGPT exhibited the weakest connectivity, recalled their own text less accurately, and reported a lower sense of ownership over what they had written.
The researchers call this phenomenon "cognitive debt", short-term relief that exacts a long-term cost.
Interesting: When participants first wrote on their own and only then gained access to AI, their brain activity remained high. The order therefore matters — first the skill, then the tool.
How AI Is Changing Learning
AI enables faster access to knowledge and personalised explanations tailored to each individual, which is a significant advantage, particularly with complex topics.
The cost is what is known as cognitive offloading: when the tool performs the mental effort on our behalf, the brain simply does not get to practise it.
The solution is not to disconnect from technology, but to use it in the right way. AI as a conversation partner for building understanding is beneficial; AI as a shortcut to a ready-made answer gradually weakens one's own thinking over time.
How AI Is Changing Decision-Making
A study by Microsoft Research and Carnegie Mellon University (Lee et al., 2025), involving 319 office workers and nearly 1,000 instances of AI use at work, revealed a clear pattern: the more someone trusted AI, the less they verified its answers.
Those with high confidence in their own professional expertise, however, consistently checked and supplemented AI suggestions.
A similar phenomenon,
automation bias, was confirmed in the medical field as well. In a German study (Kücking et al., 2024), 210 healthcare workers diagnosed complications with the assistance of an AI system that deliberately offered incorrect advice in half of the cases. Less experienced participants followed it blindly more often, while experienced physicians were quicker to detect the error.
Good news: An international study with 9,000 participants (Horowitz and Kahn, 2024) shows that we are naturally more critical of AI suggestions when it comes to high-stakes decisions. The problem lies primarily with routine, "low-importance" decisions — where attention is quickest to lapse.
How to Keep Your Mind Sharp While Using AI
- Try on your own first, then check with AI — the order matters, as it directly affects how much the brain actually learns.
- Always verify the answer, even when it seems convincing — especially when it comes to important decisions.
- Regularly practise "slow" thinking without assistance — reading, writing by hand, solving problems mentally.
- Be aware of your own level of trust in AI — the higher it is, the less critical we tend to be, as research shows.
- Use AI as a conversation partner, not as an authority to be followed without reflection.
Frequently Asked Questions About Artificial Intelligence and the Brain
Does artificial intelligence really damage the brain?Not in the sense of permanent harm, however, research does point clearly to a change in activity patterns. Studies (e.g. MIT Media Lab, 2025) record reduced activity in the regions responsible for memory and reasoning with frequent AI use, which over time can weaken independent thinking if it is not balanced with one's own mental effort.
Is using AI for learning harmful?Not necessarily, it depends on how it is used. AI that helps us understand material, which we then work through on our own, strengthens learning. AI that simply provides a final answer without any independent reflection on our part weakens it.
What is cognitive debt?A term coined by MIT researchers, it describes the phenomenon whereby the short-term comfort of using AI - faster results, less effort it leads in the long run to a weakened ability to solve problems independently and poorer memory of one's own work.
Why do people too readily trust AI suggestions?The phenomenon is known as automation bias. The tendency to accept a suggestion that appears "objective" and confident without verifying it. Research shows that the bias is more pronounced among less experienced users and in routine, low-risk decisions.
Which professional or age group is most at risk?Current research (e.g. Kücking et al., 2024) indicates that automation bias is more prevalent among less experienced users and those without specialised training in evaluating AI suggestions, regardless of age.
Summary
Artificial intelligence is undoubtedly changing the way we think, learn, and make decisions and today this is confirmed by concrete research, not merely speculation. The question is not whether AI is changing us, but how we will direct that change. Thoughtful, conscious use can turn AI into an ally of our minds, rather than a replacement for them.
Source:
Kosmyna, N. in sod. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. MIT Media Lab, arXiv:2506.08872 (predobjava). media.mit.edu
Lee, H.-P. in sod. (2025). The Impact of Generative AI on Critical Thinking. Microsoft Research & Carnegie Mellon University, CHI 2025. microsoft.com
Kücking, F. in sod. (2024). Automation Bias in AI-Decision Support. Studies in Health Technology and Informatics, 317, GMDS 2024.
Horowitz, M. C. in Kahn, L. (2024). Bending the Automation Bias Curve. International Studies Quarterly, 68(2). doi.org/10.1093/isq/sqae020