Your Brain on AI

At MIT, they've done a preliminary study of 54 participants using AI. Want to know what happens to your brain when using AI for a writing task? Here we go.

From the abstract, the researchers explored the "neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools)."

Let's break that down.

LLM stands for Large Language Models. According to IBM, AI uses LLM's "trained on immense amounts of data making them capable of understanding and generating natural language and other types of content to perform a wide range of tasks: text generation, content summarization, code generation, and sentiment analysis."

Think of it like hiring a person to write your college term paper for you, or to create Cliff Notes. While on the one hand that can seem efficient, and maybe produce something better written than you might have done, there's a serious downside.

From the research: using chatGPT "scaled down" the user's cognitive function.

EEGs are used to detect brain activity. The "EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels."

In other words, continued use of AI for language tasks makes you dumber. Significantly dumber.

That makes sense.

When I write, I have to think of what I want to say and then work to organize my thoughts so that I can effectively communicate my thoughts to someone else.

If I hire someone to do that for me or get chatGPT to do it, I'm not using my brain. I miss the workout, so to speak. My brain gets weaker, more tired, and lazy.

Computers, when done right, allow us to enhance what we already do with our brains. A computer can coalesce data faster, create a chart, and so on. But if I don't really understand the data, the argument, the presentation, and the explanation of what a computer produces for me, then it's all somewhat meaningless - because to me it has no meaning. Someone else did all of the work.

Plus, I'm a programmer. I understand completely that any computer system - no matter what it is - only does what it is programmed to do.

This is why X's grok is wildly wrong lately.

Garbage in, garbage out.

Give your brain the workout. Lay off the AI. Let's just say that there is no "AI" in a robust brAIn. Your brain is better without it.

5 Comments

Phil Waite (Jun 19, 2025 12:31 PM):

Excellent analysis!

Kim Spain (Jun 19, 2025 4:48 PM):

I'll continue to use my brain. The risk is that I may not sound as smart as I would if I used AI. That's basically a pride/ego issue that many people struggle with anyway. I'll keep struggling, but now for 2 reasons: 1) to not be prideful 2) to keep using my brain cells.

Brett Rogers (Jun 19, 2025 8:03 PM):

Agreed, Kim. AI doesn't make us better.

Randy Reeves (Jun 20, 2025 8:25 AM):

This subject can almost be tied to the political arena as well.
Using AI to say for you what you refuse to, or choose to be lazy about in gathering information, is akin to how a disturbing portion of voters, vote. Doing your own research not only rewards you in your decision to vote it arms you with information for others, with facts, that may be overlooked by others.
Our cravings for, wait for it, ARTIFICIAL INTELLIGENCE, should scare the holy crap out of us. We may not know everything you could load into a computer over many years but that does not disqualify the brain as flawed. Collectively we can recall every single bit of information.
Committing to memory, somehow, worked for 6000 years to bring history to the year 2025. I see no reason to abandon a reliable source for one that, with the pushing of a button, can erase truths.
No thanks.

Walter (Jul 4, 2025 5:59 AM):

Here is my take on the subject.

One who analyzes data can adjust the data to reflect the view point they want. So if the control is click, tell it to write X, and you’re done. That’s a simple request.

Now if I need to reason an issue, with a solid prompt. Then work out the issue over time. That’s going to produce a different result.

The majority of folks only single task AI. They need to explorer prompts to go deeper into its reasoning and logic.

My two cents.