A new study by researchers at the Massachusetts Institute of Technology (MIT) has raised concerns about the long-term cognitive impact of relying on artificial intelligence tools like ChatGPT. The findings suggest that repeated use of large language models (LLMs) may impair memory retention, reduce critical thinking, and lead to “shallow or biased perspectives.”
Conducted by the MIT Media Lab, the study divided participants into three groups and asked each to write essays using either ChatGPT, a standard search engine, or no digital tools at all. While participants worked on their tasks, researchers monitored brain activity using electroencephalogram (EEG) scans to assess neural engagement.
The group that used no tools—relying solely on their own knowledge—demonstrated the highest levels of brain activity, cognitive engagement, and memory retention. In contrast, the ChatGPT-only group showed the lowest neural activation. Participants in this group also struggled to recall or recognise their own writing, according to the researchers.
In a follow-up session, the ChatGPT group was asked to complete a similar essay task without any assistance. Those participants performed significantly worse than their peers, producing writing that evaluators described as “biased and superficial.” Moreover, they exhibited weaker connections to their own work and were often unable to recall specific quotes from their essays.
The study concluded that frequent reliance on AI writing tools could result in what researchers call “cognitive debt,” reducing users’ ability to think independently over time. The authors warned that this may make individuals more vulnerable to manipulation, weaken their capacity for critical inquiry, and ultimately diminish learning skills.
“When participants reproduce suggestions without evaluating their accuracy or relevance, they not only forfeit ownership of the ideas but also risk internalising shallow or biased perspectives,” the report stated.
The findings raise broader questions about the role of AI tools in education and human cognition. While AI platforms like ChatGPT offer convenience and efficiency, the study suggests these benefits could come at a significant cost to the user’s intellectual development.
The authors recommend further research into how AI tools affect human cognition before large language models are widely embraced as educational or productivity solutions.
The study is currently in pre-print and has not yet been peer-reviewed. However, its implications are already sparking debate over the responsible use of AI in academic, professional, and personal settings.
