NEWYou can now listen to Fox News articles!
Using ChatGPT on a long-term basis could have negative effects on brain function.
That’s according to a study led by the Massachusetts Institute of Technology (MIT), which found that using a large language model (LLM) to write multiple essays over a four-month period could hamper cognitive abilities.
In the study, 54 participants were divided into three groups.
WOMAN SAYS CHATGPT SAVED HER LIFE BY HELPING DETECT CANCER, WHICH DOCTORS MISSED
One group used ChatGPT, an LLM product made by OpenAI, to write an essay.
The second group used only a search engine, and the third group used only their own brains, according to a press release from MIT.
The participants underwent three sessions where they completed the same assignment. Then, in the fourth session, the LLM group was asked to write an essay without any tools, and the “brain-only” group was asked to use an LLM for assistance.
During each session, the researchers recorded the participants’ brain activity using an EEG monitor to assess their “cognitive engagement and cognitive load” and to determine their neural activity, the release stated.
BRAIN IMPLANT BREAKTHROUGH ALLOWS PARALYZED PATIENTS TO ‘SPEAK’ WITH THEIR THOUGHTS
The participants also provided their own individual feedback during interviews.
Human teachers and an artificial intelligence agent scored the assessments.
“EEG analysis presented robust evidence that LLM, search engine and brain-only groups had significantly different neural connectivity patterns, reflecting divergent cognitive strategies,” the researchers wrote.
Participants showed less brain connectivity when they used the tools to help write their essays, the study found.
“The brain‑only group exhibited the strongest, widest‑ranging networks; the search engine group showed intermediate engagement; and LLM assistance elicited the weakest overall coupling,” the researchers wrote.
“The key isn’t avoiding AI — it’s building cognitive strength first.”
In the fourth session, the participants who switched from LLM to brain-only showed “weaker neural connectivity” and less cognitive engagement.
The LLM group also had less ability to recall information from the essays they had just written.
Those who switched from brain-only to LLM had “higher memory recall” and greater cognitive engagement.
Based on these findings, the researchers said there could be a “possible decrease in learning skills” among LLM users.
“The use of LLM had a measurable impact on our participants, and while the benefits were initially apparent, as we demonstrated over the course of four sessions … the LLM group’s participants performed worse than their counterparts in the brain-only group at all levels: neural, linguistic [and] scoring,” they wrote.
Potential limitations
The findings have been uploaded to Arxiv, a preprint service, but have not yet been peer-reviewed, as the researchers noted that “all conclusions are to be treated with caution and as preliminary.”
There were also a limited number of participants who were all from the same geographical area.
AI TOOL SCANS FACES TO PREDICT BIOLOGICAL AGE AND CANCER SURVIVAL
“For future work, it will be important to include a larger number of participants coming from diverse backgrounds, like professionals in different areas and age groups, as well as ensure that the study is more gender-balanced,” the researchers noted.
Only ChatGPT was used in the study; future research could incorporate other LLMs.
The EEG technology used to analyze brain connectivity could also have some limitations, as the researchers shared plans to use fMRI (functional magnetic resonance imaging) in future studies.
“Our findings are context-dependent and are focused on writing an essay in an educational setting and may not generalize across tasks,” they also stated.
“Future studies should also consider exploring longitudinal impacts of tool usage on memory retention, creativity and writing fluency.”
Dr. Harvey Castro, an ER physician and “AI futurist” based in Texas, said he sees this study as a “neuro-wake-up call,” especially for younger brains.
“ChatGPT can make you 60% faster, but that speed comes at the price of neuro-engagement,” Castro, who was not involved in the study, told Fox News Digital.
“Brain connectivity collapses from 79 neural links to just 42, and 83% of users can’t quote their own essays minutes later. Neuroplasticity research tells us developing brains will feel this hit hardest.”
In emergency medicine, Castro said, doctors call this “failure to encode.”
“The brain isn’t processing and storing information,” he said. “When neural connectivity drops by nearly half, we’re looking at what researchers call ‘cognitive debt.’”
For medical students, an inability to encode and recall information under pressure could have serious implications for clinical decision-making, Castro noted.
CLICK HERE TO SIGN UP FOR OUR HEALTH NEWSLETTER
“The same neural networks that consolidate essay information are involved in diagnostic reasoning,” he said.
Using LLMs for extended periods can be convenient, but could cause cognitive muscles to “atrophy” over time, the expert cautioned.
There was one encouraging finding, however.
“When people with strong foundational skills later used ChatGPT, they showed enhanced connectivity,” Castro said. “The key isn’t avoiding AI — it’s building cognitive strength first.”
In education, he emphasized the need for periods of “AI-free cognitive development.”
For more Health articles, visit www.foxnews.com/health
“Sometimes you act on preliminary data when the stakes are high enough, and an entire generation’s brain development is high stakes.”
Fox News Digital reached out to OpenAI for comment.
Read the full article here
