Is AI Making You Dumber? Three Ways to Prevent Cognitive Decline

When AI Thinks for You, Your Brain Quietly Atrophies

Have you ever asked Claude a question, read its answer, and somehow felt more confused than before? Or maybe you've gotten so used to having ChatGPT draft your emails that when you have to write one yourself, you draw a complete blank?

You're not alone. Almost everyone who uses AI regularly runs into this. The good news: it's entirely avoidable.

Tools like Claude and ChatGPT make it dangerously easy to "outsource" your thinking — a phenomenon psychologists call cognitive offloading. We do this with mundane tasks all the time, handing off the routine stuff to free up mental bandwidth for harder problems. But when you offload thinking to AI, you're giving away far more than you realize.

Over time, core abilities like critical thinking and writing quietly erode. This is what researchers call de-skilling.

The concept of cognitive offloading was first developed by cognitive scientist Andy Clark and others. It describes how humans transfer mental tasks to external tools or environments — from writing to-do lists to using GPS navigation. Moderate cognitive offloading is harmless; it's actually a key strategy humans use to extend their cognitive reach. The problem lies in what you offload. When you hand a skill you're supposed to be developing to a tool, the neural circuits tied to that skill weaken from disuse — a classic case of the neuroscientific "use it or lose it" principle. What makes AI uniquely risky is its incredibly low barrier to entry and sweeping scope: it can substitute for nearly the entire chain of human thought, from language and expression to logic and reasoning, enabling cognitive offloading on a scale no previous tool has ever matched.

What the Experts Say: Decline Is Preventable — and Reversible

The author of this piece spoke with multiple experts — psychologists, computer scientists, and designers. Their conclusion is worth sitting with: this kind of cognitive decline is preventable, and it all comes down to how you use AI. Even if it's already happening, taking a break from AI for a while means it is reversible.

In other words, the problem isn't AI itself. It's how you use it. Here are three practical ways to use it better.

Tip #1: Be Intentional About What You Ask AI to Do

Before turning to AI, ask yourself one question: do I want to learn this, or do I just want to get it done?

If you're a beginner in a field — say, you're just starting to learn programming — and you actually want to improve, don't outsource the thinking to AI. The moment you hand it off, you lose the chance to build the skill yourself. In those moments, struggling through it with your own brain is exactly what growth requires.

On the flip side, if you're already an expert, the calculus changes. Experts note that well-established skills are more durable and don't erode as easily from AI use. For them, using AI to speed through repetitive work or handle boilerplate tasks is a genuinely efficient choice.

The core logic is simple: AI is best suited to save you time on skills you've already mastered — not to replace skills you haven't yet built.

Psychological research divides skill acquisition into two stages: "automaticity" and "controlled processing." Beginners are in the controlled processing stage, where every step demands significant conscious cognitive effort. That effortful feeling is exactly what builds stable knowledge structures in the brain. Skip that stage by jumping straight to external answers, and knowledge stays shallow — it never gets internalized into a transferable skill. Experts are different. Their core skills are already highly automated, so occasionally delegating repetitive tasks to a tool doesn't destabilize their existing cognitive architecture. This explains why the same AI tool feels like a "productivity boost" to an expert but feels like "I'm getting worse the more I use it" to a beginner.

Tip #2: Tell AI to Treat You Like a Student

If your goal is to learn something new — or to keep existing skills from fading — try actively reshaping how you interact with AI.

Here's how: update your "custom instructions" in Claude or ChatGPT and explicitly tell it that you want to learn and build skills. Ask it to follow up with questions, or even quiz you.

Don't worry about burdening the AI — it doesn't mind the extra work. This transforms the conversation from a one-way "give me the answer" exchange into a two-way "guide my thinking" dialogue. Instead of a machine that just feeds you conclusions, AI becomes something more like a tutor who pushes back and tests you. That kind of interaction, by itself, forces your brain to stay engaged.

Tip #3: Slow Down and Embrace the Friction

The last tip is perhaps the most counterintuitive: deliberately create some resistance for yourself.

A lot of people use AI by firing off a dozen prompts in a row, then receiving a wall of text they never actually read carefully. That kind of "efficiency" is mostly just spinning your wheels.

The better approach is to slow down: step away from the AI tool and go read the original sources and primary materials. That sense of friction might feel like it's slowing you down, but it's precisely what helps you actually learn things and keep your mind sharp. Learning was never meant to be frictionless coasting — it's a climb, and the resistance is the point.

Educational psychology has a concept for this called desirable difficulties, introduced by psychologist Robert Bjork: moderate resistance and challenge during learning significantly improves long-term retention and the ability to transfer knowledge. Spaced repetition, active recall, and interleaved practice all fall into this category of "productive friction." By contrast, AI's instant, complete answers eliminate all friction — leaving you feeling like you learned something when you really only skimmed it. Reading primary sources, taking notes by hand, trying to explain something in your own words — these seemingly inefficient habits are precisely the steps that trigger deep encoding.

The Ultimate Test: Go AI-Free for a Week or Two

Want to know how much AI has taken over your thinking? There's a simple way to find out: go completely AI-free for one to two weeks and rely entirely on your own brain to think and express yourself — then see how it feels.

As the original piece puts it: "If you don't go looking, you'll never know what you've lost."

That line captures something easy to overlook in the age of AI. The more powerful our tools become, the more intentional we need to be about protecting the core human capabilities that are ours alone. AI can be a lever that amplifies your abilities — or an environment that quietly erodes them. The difference comes down to one thing: whether you're still using your own brain.