Is ChatGPT Making Us Dumber? The Honest Answer

The headlines are everywhere. "ChatGPT is dramatically weakening our brains." "MIT study says yes, AI is dumbing us down." Most of them trace back to one 2025 MIT Media Lab study that hooked people up to EEG sensors while they wrote essays with and without ChatGPT, and found something researchers labeled "cognitive debt." That phrase went viral. What did not go as viral: the MIT team itself explicitly told journalists not to use words like "dumb," "stupid," or "brain damage" to describe their findings. So what is actually going on? Here is the honest answer, based on what the research actually says, not what the headlines turned it into.

What the MIT Study Actually Found

The study, formally titled "Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task," was led by researcher Nataliya Kosmyna and published as a preprint in June 2025. Fifty four participants from MIT, Harvard, Wellesley, Tufts, and Northeastern were split into three groups and asked to write SAT-style essays across several sessions while wearing 32-channel EEG headsets: one group used ChatGPT, one used a search engine, and one wrote entirely unaided.

The results were consistent across three separate measures. On EEG, the ChatGPT group showed the weakest and least distributed brain connectivity of the three groups, while the unaided writers showed the strongest. Behaviorally, ChatGPT users struggled more to accurately quote sentences from essays they had just written themselves. And on a self-reported measure of ownership, the group that wrote unaided felt the most ownership over their essays, while the ChatGPT group felt the least. Researchers describe cognitive debt as the idea that leaning on an AI tool saves mental effort in the moment but can come at a longer-term cost to memory, independent reasoning, and how deeply information actually gets processed.

The Part Most Headlines Left Out

Here is where the story gets more careful than "AI is making us stupid." First, weaker EEG connectivity is not the same thing as brain damage or permanent decline. It reflects reduced cognitive engagement during that specific task, not proof of lasting harm. The MIT Media Lab published an explicit FAQ addressing exactly this confusion, stating plainly that it is not accurate to say the study shows LLMs are making people "dumber," and asking that language like "brain damage," "brain rot," and "collapse" not be used to describe the results.

Second, the study found something researchers consider genuinely encouraging: order matters. Participants who wrote an essay unaided first, then used ChatGPT afterward, showed stronger memory recall and more widespread brain activity than participants who started with ChatGPT and then had to write alone. In other words, the same tool produced very different outcomes depending on whether it was used to think for the writer or to build on thinking the writer had already done.

Third, this was a small, short-term lab study, 54 people, writing SAT-style essays, over roughly four months, using one specific tool for one specific task. The researchers themselves flagged that their findings are context-dependent, may not generalize beyond essay writing in an educational setting, and that longitudinal research on memory, creativity, and fluency is still needed. That is a meaningfully different claim than "ChatGPT rewires your brain for the worse."

What Other Research Shows

The MIT study is not the only evidence out there, and the wider research picture is more mixed than the viral headlines suggest. A 2026 systematic review in Thinking Skills and Creativity synthesizing 67 empirical studies on ChatGPT and university students found that outcomes depended heavily on how the tool was used. When students engaged critically with ChatGPT's output, questioning it, fact-checking it, using it to generate counterarguments, several studies reported real gains in argumentative reasoning and error detection. One case study comparing college students against ChatGPT directly in live debates found that students' critical thinking actually improved over the course of the debate, and that ChatGPT was most useful as a source of counterarguments the students then had to evaluate themselves.

The pattern across this research is consistent: AI tools tend to support thinking when a person stays actively engaged with the output (evaluating it, challenging it, revising it) and tend to erode thinking when a person just accepts what the model produced and moves on. That distinction, active use versus passive acceptance, matters more than which specific tool someone uses.

So, Is ChatGPT Making Us Dumber?

The honest answer is: not automatically, and not for everyone, but it can, depending entirely on how you use it. The strongest evidence available shows that outsourcing the entire thinking process to ChatGPT, skipping the drafting, skipping the struggle, just prompting and pasting, is associated with weaker engagement, weaker recall, and less ownership over the resulting work. But the same research also shows that using ChatGPT after you have already done your own thinking, or using it as something to argue with rather than defer to, does not carry the same cost and can even sharpen reasoning.

This mirrors a debate that is not actually new. Similar concerns were raised about calculators, search engines, and GPS before ChatGPT. In each case, the tools that replaced effortful cognitive work entirely tended to reduce a specific skill (mental arithmetic, memorization, navigation), while tools used as an aid alongside active thinking generally did not. ChatGPT is a more powerful version of that same tradeoff, applied to reasoning and writing rather than arithmetic or memory.

How to Use ChatGPT Without Losing Your Edge

Based on what the research actually points to, a few habits make the biggest difference.

Think first, prompt second. Draft your own outline or argument before opening ChatGPT. The MIT study's clearest positive finding was that this order preserved memory and engagement far better than starting with the AI.
Use it to argue with, not agree with. Ask ChatGPT to challenge your position or find holes in your reasoning, rather than asking it to write the reasoning for you.
Never skip verification. Treat outputs as a draft to check, not a finished answer, especially for anything factual.
Notice when you're just copying and pasting. That specific pattern, prompt, copy, paste, no edits, is what researchers flagged as most closely tied to weaker outcomes.
Reserve some tasks for yourself entirely. Keeping certain thinking tasks, journaling, first drafts, problem-solving, fully unaided helps maintain the underlying skill even as you use AI elsewhere.

Quick Recap

The viral "ChatGPT makes you dumber" claim traces to an MIT study that found weaker brain connectivity and recall in ChatGPT users writing essays, but the researchers explicitly rejected the word "dumber" to describe it.
The same study found that writing unaided first, then using ChatGPT, preserved memory and engagement far better than starting with the AI.
Broader research on 67 studies shows mixed results: active, critical engagement with ChatGPT's output tends to support reasoning, while passive copy-paste use tends to weaken it.
The honest answer is that ChatGPT can make you dumber, but only if you let it think for you instead of with you.

Frequently Asked Questions

Does ChatGPT make you dumber?

Not automatically. Research shows the effect depends on how the tool is used. Passive use, prompting and pasting without engagement, is linked to weaker memory and reasoning, while active use, thinking first and using ChatGPT to challenge or refine that thinking, is not.

What is cognitive debt from ChatGPT?

Cognitive debt is a term coined by MIT researchers to describe how relying on an AI assistant saves mental effort in the short term but can come at a longer-term cost to memory, independent reasoning, and depth of processing.

What did the MIT ChatGPT study actually find?

Using EEG on 54 participants writing essays, MIT found ChatGPT users showed the weakest brain connectivity, struggled more to recall their own writing, and reported the lowest sense of ownership over their essays, compared to those who wrote unaided or used a search engine.

Can ChatGPT improve critical thinking instead of hurting it?

Yes, under the right conditions. Studies show that using ChatGPT to generate counterarguments, checking its answers, and staying actively engaged with its output can strengthen argumentation and error-detection skills rather than weaken them.

How can I use ChatGPT without losing critical thinking skills?

Draft your own thinking before prompting, use ChatGPT to challenge rather than replace your reasoning, verify anything factual, avoid copy-pasting outputs without review, and keep some tasks fully unaided to maintain the underlying skill.

Recommended Blogs

Why AI Gives Wrong Answers (And How to Fix It)

How to Use AI at Work in 2026: The Smart Professional's Guide

ChatGPT vs Claude vs Gemini 2026: Which AI Chatbot Should You Use?

10 AI Tools Every Professional Should Know in 2026

Learn AI in 5 Minutes a Day

Keeping up with AI does not require reading every release note yourself. Unrot delivers the day's biggest AI developments in a 5 minute daily lesson, written in plain English for beginners, students, and working professionals who want to stay current without the jargon.