Why Using AI Might Actually Ruin Your Ability to Think
I've noticed a pattern in my own AI use that is easy to miss. The moment a difficult task appears, opening an AI chatbot can feel like the obvious productivity move.
And it often works. The answer arrives faster, the blank page disappears and the task suddenly feels manageable.
But there is a second question that matters just as much: what happens to the ability you would have built if you had struggled with the problem yourself?
The biggest AI productivity question may be what happens to human reasoning when difficult mental work becomes instantly outsourceable.
The AI Productivity Paradox
The popular story about AI is simple: automate repetitive work, save time and use the extra time for more important things.
Real workplace data is more complicated. ActivTrak analyzed more than 443 million hours of activity across 1,111 organizations and 163,638 employees and found that AI adoption coincided with a more intense work environment.
Among a subset of 10,584 employees tracked before and after AI adoption, time spent on email increased 104%, chat and messaging increased 145% and business-management activity increased 94%.
Meanwhile, focused, uninterrupted work declined 9% among AI users. The workday became more productive, but also more fragmented.
ActivTrak reports these before-and-after changes for a subset of employees. The bars are scaled visually and do not represent a single common percentage axis.
David Brooks' Three Types of AI Users
In The People Who Will Thrive in the AI Age, Atlantic columnist David Brooks argues that the future divide may not be between smart and less-smart people. It may be between people who use AI to reduce mental effort and people who use it to increase their own capability.
Productive Passengers
These users hand difficult thinking to AI. They can become highly productive in the short term because the machine removes friction, but Brooks argues that repeated outsourcing can weaken the underlying skills they would otherwise develop.
Reluctant Optimizers
This group understands the danger. They intend to stay independent, but deadlines, workload and convenience repeatedly pull them toward the fastest possible answer.
Mental Marathoners
These users actively seek difficult problems. They use AI as a challenger, researcher, tutor or collaborator while preserving the parts of the work that build judgment and personal expertise.
Brooks' broader argument is that volition and a willingness to engage with difficulty become more valuable when intelligence itself becomes abundant.
The Research Behind the Concern Is Getting Harder to Ignore
One of the more important studies is a 2026 randomized set of experiments by researchers from Carnegie Mellon, Oxford, MIT and UCLA. Across three experiments involving 1,222 participants, AI assistance improved short-term performance but reduced persistence and impaired unassisted performance after the AI was removed.
The researchers found that the effect could emerge after only about 10 minutes of AI-assisted problem solving. Participants who became accustomed to immediate AI support were more likely to struggle or give up when they subsequently had to work alone.
That does not mean ten minutes of chatbot use permanently damages someone's intelligence. It means the study provides causal evidence that the way AI assistance is used can affect persistence on later unaided tasks.
Why persistence matters
Difficult thinking often feels unproductive immediately. You spend time being confused before the solution appears, and that struggle is part of learning rather than evidence that the task is failing.
The MIT Study Adds a Different Piece
A separate MIT Media Lab study examined neural and behavioral effects of using ChatGPT for essay writing. Fifty-four participants completed multiple writing sessions across AI-assisted, search-assisted and brain-only conditions.
The researchers reported differences in brain connectivity and found that the AI-assisted group showed weaker engagement on some measures. The researchers also reported differences in recall and sense of ownership of the resulting work.
This is a small study and was released as a preprint, so it should not be treated as proof that AI is "damaging the brain." Its useful contribution is narrower: it raises a testable question about whether repeatedly outsourcing composition changes how deeply people engage with the material.
There Is Evidence From Professionals Too
The concern is not limited to students. A multicentre observational study of endoscopists found that adenoma detection during non-AI-assisted colonoscopy fell from 28.4% before routine AI exposure to 22.4% after exposure.
The study does not prove that AI caused the decline; it is observational and may contain confounding factors. But the result is important because it shows why experts are beginning to ask whether AI assistance can create skill-retention problems in professional settings.
The most interesting question is not whether AI makes experts worse. It is when, how and under what workflow design reliance on AI changes the human skills that remain important when the machine is unavailable.
Why AI Can Quietly Change Your Relationship With Difficulty
Humans tend to prefer easy solutions when the stakes are low and time is scarce. A chatbot makes the easiest path extraordinarily attractive.
That can create a subtle behavioral loop: problem → prompt → immediate answer → relief.
Repeat that loop often enough and the uncomfortable part of thinking can begin to feel unnecessary. The risk is not laziness in the simplistic sense; it is that patience, uncertainty and productive struggle become less familiar.
The Atlantic article frames this as a shift from a cultivation mindset toward an optimization mindset. One is willing to endure difficulty to build capability; the other focuses primarily on getting the desired output efficiently.
The Better Way to Use AI
Ask for hints before answers
When you're learning, ask the model to identify the next step, point out an error or offer a clue. This preserves more of the reasoning process.
Start on a blank page
Write your own outline, hypothesis or solution first. Then ask AI to attack it, improve it or identify what you missed.
Use AI as a critic
Instead of saying "solve this," try: "Here is my solution. Find the weakest assumption." That turns AI into a challenge mechanism rather than a substitute thinker.
Rotate between assisted and unassisted work
If every difficult task becomes AI-assisted, you lose the chance to test whether the underlying skill is still there. Deliberately keeping some tasks AI-free creates that feedback loop.
Ask AI for thinkers, not thinking
Use AI to find relevant researchers, books, competing theories and historical arguments. Then do the deeper reading yourself.
A Simple AI Rule
- Rote: automate freely.
- Learning: ask for hints and explanations.
- Creative judgment: create your own first draft.
- Important decisions: challenge the AI and verify independently.
- Skill building: keep regular AI-free practice.
The Hidden Advantage Belongs to People Who Can Still Struggle
There is a strange possibility here. As AI becomes better at producing answers, the ability to remain comfortable while you do not yet know the answer may become more valuable.
Curiosity requires uncertainty. Creativity often begins with incomplete ideas. Deep expertise is built through repeated encounters with problems that initially resist you.
AI can make those processes faster, but it should not erase them. The goal is not to reject automation.
The goal is to make sure the machine expands your range rather than quietly shrinking it.
Pros and Cons of Heavy AI Assistance
What AI Can Improve
- Speed of completing routine work.
- Access to explanations and background information.
- Brainstorming and idea generation.
- Research and information synthesis.
- Ability to handle more tasks with limited time.
What Can Be Lost
- Persistence on difficult problems.
- Independent problem-solving practice.
- Deep engagement with source material.
- Confidence when AI assistance disappears.
- Opportunities to develop judgment through struggle.
Reclaim Deep Thinking With an E-Ink Tablet
Protect your critical thinking from AI deskilling by creating a space for uninterrupted deep work. Distraction-free e-ink tablets let you start with a blank page, outline your own ideas, and practice unaided problem-solving away from chatbots. Explore top-rated e-ink notebooks on Amazon.
Check E-Ink Tablets on Amazon →The Bottom Line
The biggest AI mistake may be thinking productivity and capability are the same thing. They are not.
AI can make you dramatically faster today while giving you fewer opportunities to practice the skills you'll need tomorrow. Current research does not prove that everyone who uses AI becomes less capable, but it does show enough warning signs to make passive reliance a poor long-term strategy.
The smartest approach is not to abandon AI. It is to become much more deliberate about which parts of thinking you delegate and which parts you keep for yourself.
Let AI handle repetitive work. Let it challenge your assumptions. Let it find information and expose you to ideas you might have missed.
But keep some blank pages. Keep some difficult problems. Keep some moments when there is no instant answer.
Because in an age where intelligence is increasingly available on demand, the people who continue to develop their own judgment, curiosity and persistence may have the advantage that no model can simply hand them.
Deskilling Isn't the Only AI Risk in 2026
While cognitive offloading quietly erodes human critical thinking, a much louder threat is emerging across automated infrastructure. Beyond simple chatbots, what happens when multi-agent workflows are intentionally weaponized? Read our complete deep-dive report into the real, overlooked AI security threats of 2026.
Read the AI Threats Deep-Dive →Sources
The Atlantic: The People Who Will Thrive in the AI Age
ActivTrak: 2026 State of the Workplace
MIT Media Lab: Your Brain on ChatGPT
arXiv: AI Assistance Reduces Persistence and Hurts Independent Performance
Carnegie Mellon University: Human-AI Research Workshop
Nature Reviews Gastroenterology & Hepatology: AI-assisted colonoscopy and risk of endoscopist deskilling
PubMed: Endoscopist deskilling risk after exposure to artificial intelligence
Wharton: Thinking Fast, Slow, Artificially: AI and Your Brain
Frequently Asked Questions (FAQ)
Does AI reduce critical thinking?
Not automatically. Research suggests the effect depends heavily on how AI is used: outsourcing the entire reasoning process can reduce persistence and independent performance, while using AI for feedback, hints and challenge can preserve more human engagement.
Can AI make people more productive but less capable?
Potentially. Workplace data shows AI can increase activity and productivity while focused work falls, and experiments indicate that some forms of AI assistance can impair performance when people later have to work without AI.
What is the best way to use AI without losing thinking skills?
Start difficult tasks yourself, ask AI for hints or critique, verify its reasoning, alternate between AI-assisted and unaided practice, and reserve routine work for automation.
Did the MIT ChatGPT study prove AI damages the brain?
No. The MIT study was relatively small and examined neural and behavioral differences during AI-assisted writing. It raises important questions about cognitive engagement but does not establish permanent brain damage.
What is “cognitive offloading” in AI?
Cognitive offloading means using an external tool to perform part of a mental task. With AI, this can include delegating writing, planning, remembering, analysis or problem-solving that a person could otherwise perform themselves.
No comments:
Post a Comment