A recent study from MIT suggests that using AI chatbots like OpenAI’s ChatGPT could impair cognitive function.
Researchers at the Massachusetts Institute of Technology’s Media Lab conducted an experiment involving 54 participants, who were asked to complete essay-writing tasks across three sessions. Each participant used one of three methods: ChatGPT, traditional search engines, or their own unaided reasoning.
In a fourth session, the roles were reversed—those who had relied on ChatGPT were asked to write without any tools, while the brain-only group was instructed to use a large language model (LLM).
The findings were striking: over 83% of the ChatGPT users were unable to recall or quote from the essays they had written just minutes earlier, indicating significantly poorer memory retention compared to the other groups.
Severe Cognitive Impairment
Alex Vacca, co-founder of the sales tech agency ColdIQ, called the findings “terrifying,” warning that AI isn’t enhancing our productivity — “it’s leaving us cognitively bankrupt.”
“You write something, hit save, and your brain has already forgotten it because ChatGPT did the thinking.”
The researchers observed that brain connectivity “consistently declined” as reliance on external tools increased.
“The Brain-only group demonstrated the most robust and wide-ranging neural activity, the Search Engine group showed moderate engagement, and those using LLMs exhibited the weakest overall brain connectivity,” they reported.
Brainpower on borrowed time
To assess participants’ cognitive engagement and mental effort during the tasks, the researchers measured brain activity using electroencephalography (EEG). Their findings revealed that repeated reliance on external systems like large language models (LLMs) can lead to an accumulation of “cognitive debt” — a gradual decline in independent thinking as key cognitive processes are outsourced to AI.
“Cognitive debt defers mental effort in the short term but results in long-term costs, such as diminished critical inquiry, increased vulnerability to manipulation, decreased creativity.”
Impairment of Learning Functions
The paper — which is still awaiting peer review — suggests that using AI language models (LLMs) may actually hinder learning, particularly among younger users.
“Our findings highlight an urgent concern: a potential decline in learning abilities linked to LLM use,” the researchers concluded.
They emphasized the need for “longitudinal studies” to better understand the long-term effects of AI chatbots on the human brain, cautioning against embracing these tools as universally beneficial “before LLMs are recognized as something that is net positive for humans.”
When asked by Cointelegraph about the study, ChatGPT responded, “This study doesn’t claim ChatGPT is inherently harmful — rather, it cautions against overreliance without critical thinking or personal effort.”
