MIT study indicates that A.I. usage can lead to a drop in critical thinking skills.
Photo courtesy of MIT Media Lab
By Julia Birdsall
Have you ever used A.I. to take a test, write an essay or complete any homework assignment?
If so, you’re not alone. According to the Digital Education Council, “86% of students use AI in their studies, with 54% using it weekly and nearly one in four (25%) using it daily.”
A University of South Florida health sciences major who has chosen to remain anonymous spoke with The Crow’s Nest about how they use A.I. for homework.
“I use A.I. for basically every assignment, sometimes just to check, but a lot for the classes that I don’t care as much about,” they said. “I don’t think it affects my critical thinking skills; it just makes me want to use my brain less.”
It is those students using it daily that have become a particular concern, as studies have begun to show a link between A.I. reliance and a decline in A.I. users’ critical thinking skills.
A study published by Michael Gerlich of the Swiss Business School (SBS) claims that cognitive offloading is the link.
Gerlich defines cognitive offloading as “the externalization of cognitive processes, often involving tools or external agents, such as notes, calculators, or digital tools like AI, to reduce cognitive load.”
By assessing questionnaire results from 666 participants, Gerlich was able to conclude that the participants who used A.I. showed higher levels of cognitive offloading and lower critical thinking skills.
These findings are cause for concern in schools and the workforce.
There are now many jobs in which human workers are forced to compete with A.I. for positions. How can they do that if they lack the creative and critical thinking skills required?
How will they learn those skills if A.I. use is a frequent occurrence at all levels of education?
An MIT Media Lab study may have a solution.
The author of the study, Nataliya Kosmyna, recently spoke to Andrew Chow from Times about the study’s results.
The study’s participants were asked to write essays based on SAT questions, which they did in three sessions. They were split into three groups: the “LLM (Language Learning Model) group” only used ChatGPT, while the “Search Engine group” used a non-A.I. search engine, and the “Brain-only group” used only their own knowledge.
While participants wrote, researchers monitored their brain activity using an electroencephalogram (EEG).
Kosmyna told Times that there was a difference between the brain activity between each of the groups. Less brain activity occurred in the LLM group than the brain-only group, with the search engine group serving as a middle ground between the two.
Participants were then invited back for a fourth session, in which they were switched to a different group, but they were able to pick up on a previous essay topic of their choosing.
The results of this fourth session revealed that members of the brain-only group that were switched to the LLM group had heightened performance with ChatGPT.
More research will have to be done to solidify these results, but it raises the possibility that A.I. has the possibility to become a way of increasing performance in users that work with the systems rather than using them as a crutch.
