Is AI Making Us Dumb?

There's no denying that AI is both incredibly useful and legitimately concerning.

It can summarize a 40-page PDF, draft an email, explain quantum tunneling, and help you brainstorm at 2 a.m. But how does all that offloading impact our brains in the long term? Memory, focus and critical thinking are all better with practice. What happens when we stop practicing?

Thankfully, that problem is specific, and researchers are already finding fixes.

A small study by Zeineb Farhat in Applied Cognitive Psychology interviewed 45 young adults in France, ages 18 to 25, who used ChatGPT every day. Many said that getting instant answers made them less likely to analyze problems, check sources, or wrestle with ideas on their own.

Participants reported “intellectual laziness,” shorter attention spans, and trouble remembering things. Instead of trying to recall facts from memory, they let ChatGPT do it, basically turning AI into an external memory drive.

Some also said it comes with a psychological cost… less confidence in their own judgment, more second-guessing, and even anxiety when ChatGPT was unavailable. A few replaced conversations with people by talking to the bot instead, which some linked to social isolation.

If this reminds you of the spread of the internet & smart phones, you're not wrong.

A separate MIT study followed 54 adults for four months as they wrote essays using either ChatGPT, a search engine, or no digital assistance (“brain-only”). The ChatGPT group showed lower cognitive engagement, had more difficulty recalling quotations from their own essays, and felt less ownership over what they produced.

The researchers called this cognitive debt. You get the finished product now, but you may pay later in reduced learning, weaker recall, and less practice at independent reasoning. Interestingly, only 18 participants completed the final crossover session, so the authors themselves treated the findings as preliminary, not as proof that ChatGPT permanently damages anyone’s brain.

Think of it like using a calculator. If you use one to check your work after attempting the math, you’re still exercising numerical reasoning. If you use it to avoid ever attempting the math, the skill gets rusty. AI use works the same way... except it can substitute for a much wider range of mental work.

Our current fundamental understanding is this: Relying heavily on AI will make you worse at the exact skills you give up, but not less intelligent overall. The realistic  concern is “cognitive offloading,” or letting a tool do your remembering, evaluating, writing, or deciding for you.

Offloading is fine when the task isn’t important, like formatting a spreadsheet, fixing grammar, or summarizing something you’ll double-check. It can enable you to do things that are beyond your skill set. But it gets risky when AI does the evaluating, synthesizing, or deciding, which are the exact exercises that actually build your skills.

Memory isn’t just a filing cabinet. It gets stronger when you try to remember things, explain them in your own words, or connect them to things you already know. If you always ask a chatbot right away, you skip the productive struggle that builds memory.

That’s why the interview participants’ experience makes sense: “I just ask ChatGPT whenever I need something… I don’t retain much.” The information may be available instantly, but it is not actually yours.

So, how do you make the most of a tool like AI without getting mentally flabby?

You need boundaries that keep your brain in the loop.

  • Try first, then ask. Try to answer, draft, or decide before consulting AI. Use the response to compare and improve your thinking, not replace it.
  • Make AI push back. Ask it for the strongest counterargument, missing evidence, or ways your reasoning could be off.
  • Double-check anything important. Treat AI’s answer as a rough draft, not a citation. Always check sources, especially for health, legal, money, or science stuff.
  • Practice recall. After you learn something, close the chat and try to explain it from memory, then see what you missed.
  • Keep the tough stuff for yourself. Let AI handle formatting or summarizing, but you make the judgment calls. Things like interpretation, ethics, priorities, and final decisions should never be left to a chatbot.
  • Talk to actual humans. This cannot be stressed enough! The new study’s social-isolation findings are a reminder that discussing ideas with other real people is not just pleasant; it’s a vital cognitive practice.

If a task is critical, a skill set you want to stay sharp, or something you want to get better at, it's a job for your brain.

This technology is still in its early stages. The research is the same: small, and more suggestive than conclusive
. 

Ultimately, the real question isn’t “Should I use AI?", but “HOW should I use AI?”

The smartest move? Ask AI to teach you something you can then repeat and improve on. Let it help you level up, not check out.


Sources

https://onlinelibrary.wiley.com/doi/10.1002/acp.70205

https://neurolaunch.com/intellectual-slothism/

https://www.media.mit.edu/publications/your-brain-on-chatgpt/

https://www.forbes.com/councils/forbestechcouncil/2025/11/26/cognitive-debt-the-hidden-cost-of-generative-ai/


Back to blog