I've noticed a strange pattern in the AI era: the easier a task becomes, the easier it is to stop doing any of the thinking behind it.
ChatGPT can write the email. It can explain the code. It can summarize the paper. It can even give you a polished argument before you've figured out what you actually believe.
That convenience is the heart of the current debate over whether AI is making us “dumber.”
The answer is not simply yes or no.
Research increasingly suggests that AI can improve performance while simultaneously reducing the amount of thinking people put into the task. Whether that becomes harmful depends heavily on what you delegate and what you continue to practice.
AI can remove mental effort from a task. The important question is whether it is removing unnecessary effort or the practice that builds a skill.
What Is Cognitive Offloading?
Humans have been outsourcing mental work for centuries.
We use calculators instead of doing arithmetic in our heads. We use GPS instead of memorizing every turn. We write reminders instead of relying entirely on memory.
Psychologists call this cognitive offloading: moving some mental work from your brain to an external tool.
There is nothing inherently wrong with it.
The problem appears when the outsourced task is also the mechanism through which you develop or maintain a skill.
Using a calculator does not necessarily matter if you already know how arithmetic works. Never learning arithmetic because a calculator always answers for you is a different situation.
Why AI Is Different From a Calculator
A calculator gives you a result.
Generative AI can produce the reasoning, wording, structure and ideas around that result.
That makes the cognitive handoff much larger.
That distinction is especially important for writing, coding, research, analysis and learning.
You can ask AI to critique your argument and then strengthen it yourself. Or you can ask AI to produce the entire argument and simply approve the final answer.
Both actions save time.
They do not provide the same learning experience.
What the MIT ChatGPT Study Actually Found
A 2025 MIT Media Lab study placed 54 participants into three groups: LLM-assisted writing, search-engine-assisted writing and writing without external tools.
Researchers used EEG alongside linguistic and behavioral measures while participants completed essay-writing tasks.
The LLM group showed the weakest measured brain-connectivity patterns of the three groups during the task, while the brain-only group showed the strongest and most distributed connectivity.
The researchers also reported lower self-reported ownership of the essays among the LLM group and difficulty accurately recalling or quoting their own writing.
Those findings are interesting, but there is an important limitation.
This was a small, specific essay-writing experiment. It does not prove that using ChatGPT casually damages everyone's brain or intelligence.
The researchers themselves call for deeper work on the long-term implications.
Microsoft Found Something Similar at Work
Microsoft Research surveyed 319 knowledge workers and collected 936 real-world examples of AI-assisted work.
The researchers found that higher confidence in AI was associated with less reported critical-thinking effort, while higher confidence in one's own abilities was associated with more critical-thinking effort.
They also observed something more nuanced: AI changed what critical thinking looked like.
Instead of spending all the effort producing information, workers increasingly spent effort verifying, integrating and supervising AI output.
The Shift in Cognitive Work
- Before AI: more effort goes into finding, producing and structuring the material.
- With AI: more effort can shift toward checking, editing, comparing and deciding.
- The risk: if the checking stage is skipped, the user may stop exercising the underlying skill.
That is a much more useful way to think about the AI debate than simply asking whether ChatGPT is “good” or “bad.”
A 2026 Study Adds an Important Twist
A June 2026 study of 120 young adults compared AI-assisted and manual performance on a logical-fallacy task.
The AI group produced more correct answers and reported lower mental effort.
But the researchers did not find that AI uniformly improved or reduced critical thinking.
Instead, they observed different styles of AI use, ranging from complete delegation to strategic engagement.
The Difference Is Delegation vs. Scaffolding
When AI replaces the thinking, the user gets the answer. When AI acts as a scaffold, the user still performs the reasoning and uses the model to test, explain or extend it.
That distinction may be more important than the amount of time you spend using AI.
AI Can Improve Performance Without Improving Skill
This is the paradox that causes so much confusion.
You can become better at producing a report with AI while becoming less practiced at writing reports yourself.
You can ship code faster with an AI coding assistant while becoming less fluent at constructing the same code from scratch.
Your final output improves.
Your independent ability may not.
A recent review in Trends in Cognitive Sciences makes this distinction explicit: researchers argue that offloading can impede skill acquisition and contribute to skill decay, while also noting that basic cognitive abilities may be more resilient than learned skills. The review concludes that the long-term effects remain uncertain and depend on how AI is used.
Socrates Had a Warning That Still Fits
— Socrates, as recorded by Plato in the Apology
That is not an argument against technology.
It is an argument for examination.
And that may be the simplest defense against over-reliance on AI: do not let the tool become the end of the thinking process.
Young Americans Are Already Worried About This
Pew Research Center's 2026 survey found that adults ages 18 to 29 are more skeptical about AI's long-term impact than older groups.
37% of adults ages 18 to 29 said they expected AI to have a negative effect on them personally over the next 20 years.
Pew also found that 20% of adults under 30 said chatbots hurt their creativity, compared with 25% who said they helped it.
That is not proof that AI is damaging creativity.
It does show that people who use these systems heavily are increasingly thinking about what happens when convenience begins replacing effort.
How to Use ChatGPT Without Outsourcing Your Brain
The goal does not have to be abandoning AI.
A better strategy is to deliberately protect the skills you still need to own.
Five Rules for Smarter AI Use
- Think first. Write your initial answer, outline or hypothesis before asking AI for one.
- Ask for hints. For learning tasks, request clues, questions or explanations instead of the final answer.
- Compare outputs. Ask for several alternatives and explain which one you would choose and why.
- Verify independently. Do not allow a confident AI response to become your evidence.
- Practice without AI. Regularly perform important skills from memory and without assistance.
The Most Powerful Trick Is Adding Friction
AI's greatest advantage is speed.
That is also where the cognitive risk can hide.
When an answer arrives instantly, you lose the pause during which your brain might have formed its own hypothesis.
One of the most useful suggestions from researchers cited by Vox is therefore surprisingly simple: slow down.
Ask the AI for five possibilities instead of one. Ask it to challenge your reasoning. Ask for counterarguments. Ask for questions you should be able to answer yourself.
Comparison forces judgment.
Judgment is exactly the skill you do not want to outsource.
Books Worth Keeping on Your Desk
Kindle Scribe E-Ink Notebook
If you want to stop outsourcing your thinking, start your writing process away from the keyboard. A premium e-ink tablet adds the necessary friction of handwriting, forcing you to develop your own ideas before an AI can generate them for you.
Check Kindle Scribe on Amazon →Sony WH-1000XM5 Noise Canceling Headphones
Unassisted critical thinking requires sustained focus. Industry-leading active noise cancellation helps you block out office distractions and notification fatigue so you can actually concentrate on the hard mental work.
Check Sony WH-1000XM5 on Amazon →Albert Einstein's Advice May Be the Better AI Prompt
— Albert Einstein
That is almost the perfect rule for using generative AI.
Do not stop questioning just because the machine gives you an answer.
Question the source. Question the assumptions. Question the missing information. Question whether you would still reach the same conclusion without the chatbot.
That turns AI from a vending machine into a thinking partner.
So, Can ChatGPT Make You Dumber?
There is not enough evidence to say that ChatGPT simply makes people less intelligent.
There is, however, growing evidence for something much more specific.
When AI repeatedly replaces the mental practice required to build or maintain a skill, that skill can receive less practice.
That is the real danger.
The solution is not to reject AI.
It is to decide which mental work you want AI to remove and which mental work you still want to own.
Let AI handle repetitive work.
Let it challenge your assumptions.
Let it show you alternatives you would not have considered.
But keep enough friction in the process that you still have to think.
Because the most valuable thing AI can give you is not an answer.
It is more time and more leverage to produce better thinking yourself.
Is ChatGPT Ruining Student Learning?
If AI can weaken critical thinking in professionals, what is cognitive offloading doing to students in the classroom? Read our complete 2026 guide to discover how students can safely use ChatGPT as a learning scaffold rather than an answer key that bypasses real education.
Read the Student ChatGPT Guide →Sources checked for this article:
Vox — AI can make you smarter. Here's one hack to get there.
MIT Media Lab — Your Brain on ChatGPT: Accumulation of Cognitive Debt
Microsoft Research — The Impact of Generative AI on Critical Thinking
Trends in Cognitive Sciences — Is AI Making Us Stupid?
PubMed — Cognitive Offloading, Critical Thinking and Attitudes Toward AI
Pew Research Center — Americans and AI 2026
Frequently Asked Questions About AI and Critical Thinking
Can ChatGPT actually make you dumber?
There is not enough evidence to conclude that ChatGPT simply lowers human intelligence. Research does suggest that repeatedly outsourcing tasks that require practice can reduce the practice needed to acquire or retain particular skills.
What is cognitive offloading in AI?
Cognitive offloading means using an external tool to reduce the mental effort required for a task. ChatGPT expands this concept because it can outsource not only memory or calculation but also writing, reasoning, idea generation and information synthesis.
What did the MIT ChatGPT brain study find?
The 2025 MIT Media Lab study involved 54 participants completing essay-writing tasks with an LLM, a search engine or no external tool. The researchers reported weaker measured brain connectivity in the LLM group and lower reported essay ownership, while also emphasizing the need for deeper research.
Can AI improve critical thinking instead of hurting it?
Yes. Research suggests AI can support critical thinking when people use it strategically to compare ideas, receive feedback, verify information, explore counterarguments and deepen their understanding rather than simply delegating the task.
How can I use ChatGPT without losing important skills?
Keep practicing high-value skills without AI, think through problems before asking for answers, request hints instead of solutions when learning, compare multiple AI outputs, and independently verify important claims.
No comments:
Post a Comment