What the Study Actually Found
The research comes from Michael Gerlich at SBS Swiss Business School, published in the journal Societies in early 2025. The design was mixed-method: a survey of 666 participants spanning a range of ages and education levels, followed by 50 in-depth interviews to add texture to the numbers.
The headline finding is direct. Gerlich concluded that frequent AI tool usage and critical thinking abilities show a strong negative correlation — mediated by a mechanism called cognitive offloading.
Before that sentence does too much work, one word deserves your attention: correlation. Gerlich is careful about this distinction, and so should you be. The study highlights correlations rather than direct causation. Additional factors may also contribute to the relationship, and the reverse direction is entirely plausible — people who already engage in less critical thinking may simply reach for AI tools more readily.
The study cannot tell you which way the arrow points. That matters.
The Cognitive Offloading Mechanism
Cognitive offloading is not new. It is the habit of delegating a mental task to something outside your head — a calculator, a sticky note, a search bar. We have always done it. The question the study raises is what happens when the external tool is capable enough to handle the entire job, not just store the answer.
There is a companion piece worth knowing about here, though it comes with heavier caveats. An MIT Media Lab study by Nataliya Kosmyna and colleagues had 54 participants write essays under three conditions: using an AI assistant, using a search engine, or relying solely on their own cognition — all while measuring brain activity. The brain-only writers showed the strongest neural connectivity. The AI users showed the weakest, and many struggled to recall work they had just produced.
The sample is small. The follow-on crossover phase included only 18 participants. Treat it as a suggestive signal rather than settled proof. Even so, it points at the same intuition from a different angle: when the tool does most of the assembling, less assembling is happening in the mind of the person using it.
The Age Pattern Is the Part Worth Taking Seriously
The youngest cohort in Gerlich’s study — participants aged 17 to 25 — used AI tools most frequently and scored lowest on critical thinking assessments. Those aged 46 and above used AI least and scored highest. Higher education levels were protective across all age groups.
Gerlich’s interpretation is appropriately tentative. He suggests that digital natives, who have grown up with AI-integrated technologies, may be more prone to cognitive offloading than older generations — not because they are less capable, but because the tools were always there. Reflexive reliance is easier to develop when there was never a period without the option.
That reading is softer than “young people can’t think.” It is a researcher observing that habitual access shapes habitual behavior.
Why This Matters for Knowledge Workers and Founders
If your competitive edge is judgment — reading a market, evaluating a strategy, synthesizing ambiguous information — then the cognitive muscle behind that judgment is worth protecting deliberately. Population-level patterns are not individual verdicts. But they are useful signals about where habits drift when left unexamined.
What the Research Actually Recommends
Gerlich does not land on “stop using AI.” His own conclusion is more nuanced and more useful: AI tools are not inherently detrimental. Their impact depends entirely on how they are used.
His framing is worth keeping: AI should complement cognitive engagement rather than replace it. That distinction — extension versus substitution — turns out to be the question that matters most in practice.
The difference between using AI to pressure-test your thinking and using AI to avoid thinking in the first place is not always obvious in the moment. But it is almost always obvious in the output.
Three Practical Shifts That Hold Up Against the Research
You do not need to abandon your AI stack. You need to be deliberate about where the handoff happens.
1. Keep the Judgment Layer
Outsourcing a lookup is categorically different from outsourcing the interpretation of what that lookup means. Use AI to surface information faster. Keep the evaluation of what that information means firmly in your own hands.
2. Notice When You Are Skipping the Hard Part
The hard part is usually the moment just before you open the tool. It is the moment where you would have had to sit with ambiguity, form a rough hypothesis, or make a call without certainty. That friction is not inefficiency. It is where thinking happens.
3. Audit Your Prompts Periodically
If your prompts are consistently asking AI to produce conclusions rather than to challenge yours, that is a signal worth acting on. Flip the dynamic occasionally — use the tool to stress-test a position you already hold rather than to generate one you haven’t formed yet.
The Honest Takeaway
The 2025 Gerlich study is observational. It cannot prove that AI tools cause cognitive decline, and it does not claim to. What it does establish is a meaningful correlation that deserves more than a dismissive response from people who use these tools professionally.
The tools are not the problem. Unreflective use of the tools is the problem — and that is something you can actually do something about.
The question worth sitting with is not whether you use AI. It is whether the version of you that uses AI is still doing the thinking, or whether you have quietly outsourced that part too.
Comments (0) No comments yet
Want to join this discussion? Login or Register.
No comments yet. Be the first to share your thoughts!