Skip to main content

MIT Did A Study On AI, And The Results Are Interesting

 They're probably not what you think.

Photo by Growtika on Unsplash

(This article was originally posted for paying members on Medium.)

When I began writing this article, I didn’t know anything about the Cognitive Debt AI study beyond the sentiment that, perhaps, AI is making us “dumb”. But when I started looking into it more, I realized this issue was a lot more nuanced than it seems.

Earlier this month, an group of MIT researchers published a study on the effects of using AI to help with school work.

Researchers recruited 54 greater-Boston area residents aged 18–39 to write SAT-level essays. Participants were limited to either being able to only use LLMs in their writing process, using any websites besides AI, or not being allowed to use the internet at all. In total there were 4 sessions that took place over the course of 4 months, with the last session switching the groups that subjects were originally placed into.

Researchers found that AI is affecting our brains in certain ways. Mainly, they found that neural connectivity in the brain is much weaker when subjects went from LLM-only to brain-only writing, compared to those who started in the brain-only group. 

The other thing they found that *I* thought was interesting was the affect on participant’s ability to recall or quote information that was written with AI. Those who wrote with AI assistance were unable to quote anything from their essays in the first session.

Overall, they looked at many different items of interest in their study, from what’s discussed above to even things like the perception of ownership in AI-assisted writing. 

There were many immediate and intense reactions to this research. Many of us saw major publications posting about this study using words like “dumb” and “brain rot”. 

However, this was seemingly not the original intent. 

Primary author, Nataliya Kos’myna, has done a really good job of trying to head the negative public perception of their study, even going so far as to ask journalists to avoid certain language as to not glance over the reality of their findings.



There have also been a few criticisms of the study, notably the fact that it is not yet peer-reviewed. Also, the number of participants is very small. But many people are failing to acknowledge that the others have said and are aware of these faults. 

Notably, they are currently in the process of peer review. However, researchers felt as though the findings were too important to not be discussed and receive wide feedback on. 

I really enjoy that we are coming to a point where we are examining AI very closely. I think if we as a society want to continue to use AI as a productive tool, we need to take precautions. 

This study does not make the claim that AI is harming us. What it does do is raise concern and start a conversation about the effect it has on us. And “effect” is not a dirty word. Even in the study, at first the LLM’s were seen as beneficial, but over time participants got lazy. I think this says a lot about how we should and shouldn’t try to use AI.

It’s not there to do our job for us. But, it has plenty of potential to make our jobs easier if we come together and find solutions to the current AI problems. Hopefully these solutions will do nothing but benefit us all in the end.

This article was written by Haley Marie Salgado
Find me elsewhere: X | Blogger | Substack (coming soon)

Comments

Post a Comment

Popular posts from this blog

The Best Programming Languages To Learn in 2026 (And Why They Matter)

 Whether you’re a brand new developer or shifting careers, choosing the right programming language to learn in 2025 isn’t just about popularity. It’s also about positioning yourself as someone with the right set of skills for potential future employers. Over time we have witnessed various technological improvements that have lead programming as we once knew it to change vastly. For this article, I looked at a few different sources to figure out which languages are actually in demand right now. One interesting source I discovered is called SkillQuery (this isn’t sponsored, I just thought it was neat). This is a small project created by a PhD student to show which skills are actually in demand for certain industries. 1. Python This one is a dead giveaway.  As a statistics undergrad, the need to learn Python has been absolutely drilled into my head (although I prefer working with R a lot more). But I have to admit, this is for good reason. Most sources I checked listed Python a...
Post a Comment
Read more

Is HTML Worth Learning in 2025?

If you are new to the world of coding and computers, chances are you have asked yourself this question a multitude of times for various programming languages. And although it’s not a “programming language” in the traditional sense, HTML is no different. As the tech world rapidly evolves and new and improved tools with shiny, already built-in AI features become readily available to users, it almost seems like languages such as HTML are becoming less important. However, I would argue that this isn’t actually true. While learning just HTML on it’s own won’t land you a job in tech, it can certainly strengthen your skills and provide a solid base to expand on. In this article, I will give you 4 reasons why HTML is still worth learning in 2025. 1. HTML is the foundation for the internet Nearly every website you visit relies on some form of HTML. It’s essentially the skeleton of the internet. CSS just makes it pretty, and JavaScript makes it do things. HTML is the part that gives everything a...
Post a Comment
Read more

I Was "Vibe Coding" Before It Was Even A Thing - How It Led Me To Burn Out Doing My Passion

Photo by Gustavo Zambelli on Unsplash (This article was originally posted for paid members on Medium. ) When I was 15, I discovered coding.  I was placed into a generic “Programming 1010” class in high school, and learned Scratch. Before this, I never really knew what I wanted to do as an adult. Often, people would ask me the dreaded “So, what do you wanna do when you grow up?” and I would answer with something just to please them. Oh, something with art maybe? I don’t know. Maybe a nurse. A teacher? Nothing ever stuck with me.  I was always a B average student in school. I did fine getting by, but I just didn’t have interests in the way that other people did. I grew up dealing with a pretty rough childhood, so I didn’t have time to discover what I really liked until I got older. My only hobbies at that point were listening to music and watching horror movies like the hormonal emo teenager I was (and playing Skyrim for 8 hours straight, of course). But something clicked in my...
Post a Comment
Read more