Cognitive Offloading: Does Using AI Make You Think Less?

A student finishes a history essay in ten minutes. The paragraphs are coherent, the grammar is clean, and the thesis is defensible. The student has learned almost nothing. This is the quiet worry parents and teachers everywhere are starting to voice: when AI does the thinking, does the thinker ever grow?

The question has a name in cognitive science. It is called cognitive offloading, and researchers have been studying it since long before ChatGPT existed. What is new is the scale. For the first time in human history, students can offload nearly any mental task, from arithmetic to argumentation, with a few words typed into a chat window. Understanding what that does to the developing brain is one of the most important educational questions of our decade.

What Cognitive Offloading Actually Means

Cognitive offloading is the act of using something outside your head to carry a mental load. Writing a phone number on paper so you do not have to memorize it is cognitive offloading. Using a calculator for long division is cognitive offloading. Asking a search engine where you parked your car is cognitive offloading.

Humans have been doing this forever. Writing itself is a form of offloading memory onto physical media. The invention of the printed book, the pocket calculator, and the search engine each sparked worries that people would stop thinking. In many cases those worries were overblown. We did not stop doing math when calculators arrived. We stopped memorizing phone numbers, and life went on.

What makes AI different is the type of thinking it can offload. A calculator handles computation. A search engine retrieves facts. A large language model can draft the essay, structure the argument, weigh the counterpoints, and explain itself in plain English. That is not a tool for carrying memory. That is a tool that can replace reasoning.

What the Research Shows

In 2025, Michael Gerlich of SBS Swiss Business School published one of the first large studies on AI use and critical thinking. He surveyed 666 participants across different age groups and used both self-report scales and standardized thinking assessments. The pattern was clear: participants who leaned on AI tools most heavily scored significantly lower on critical thinking measures. The relationship was strongest among younger users, who had less time to develop baseline reasoning habits before AI became available.

Follow-up reviews have replicated the direction of the effect, though with important caveats. The link between AI use and weaker thinking is a correlation, not a one-way causal arrow. It is possible that students who struggle with critical thinking reach for AI more often, rather than AI making them worse. Both are probably true at once.

The problem is not that AI gives answers. The problem is that answers alone do not build minds. Minds are built by the struggle to find answers.

Cognitive psychologists call this the desirable difficulties principle. Learning that feels easy often is not real learning. The effort you put into retrieving a fact, solving a problem, or constructing an argument is what encodes it in long-term memory and builds the neural circuits for similar problems in the future. When AI removes the struggle, it also removes the growth.

Why Young Brains Are Especially at Risk

Executive function (the mental machinery for planning, working memory, and self-control) develops into the mid-twenties. The teenage brain is literally under construction. Every time a student wrestles with a hard math problem, drafts a messy first paragraph, or forces themselves to recall a historical date, they are strengthening that construction.

Offloading those moments to AI during the formative years is a bit like skipping leg day for a decade and expecting to run a marathon. The capacity is built by the strain, not avoided by it. This matters for parents and teachers because the critical thinking a student does not practice at 14 may not simply appear at 24.

The Good News: It Depends on How AI Is Used

Here is the part the alarmist headlines miss. The same studies that found weaker thinking in heavy AI users also found something else: students who used AI to engage with material (asking it questions, debating its answers, probing weak spots) often performed better on reasoning tasks than peers who did not use AI at all.

In other words, AI is like a kitchen. You can order takeout every night and never learn to cook, or you can use the same tools to become a chef. The outcome depends on the behavior, not the technology.

What good AI use looks like

• Attempt first, ask second. Try the problem alone. Then bring your attempt, including the parts that stumped you, to AI.
• Ask for the question, not the answer. Instead of "solve this equation," try "what should I be noticing in this equation?" The second prompt teaches. The first replaces thinking.
• Make AI defend itself. Follow up any answer with "why" or "what would be wrong with this reasoning?" Treat it as a study partner you need to catch out, not an oracle you trust blindly.
• Verify everything that matters. AI can be fluent and wrong at the same time. Cross-check facts, dates, and citations against primary sources.
• Write first, polish second. Draft your own messy version of anything that matters before letting AI touch it. Your voice should be the first voice.

What bad AI use looks like

• Copy-pasting prompts straight from a worksheet and submitting the output.
• Letting AI write the first draft of anything you were supposed to think through.
• Trusting its answers without checking them.
• Using it to avoid the uncomfortable parts of learning: struggle, confusion, slow progress.

How Parents and Teachers Can Help

Rules banning AI outright tend to fail. Students use it anyway, just more secretly, and they lose the chance to learn how to use it well. A better strategy is to build what educators call AI literacy: the habits that let a student benefit from the tool without being hollowed out by it.

Start by naming the trade-off out loud. Explain that any mental task you hand off is a muscle you are not training. Then help the student decide, task by task, which muscles they want to keep strong. Mental math? Probably worth doing by hand. Formatting a bibliography? Probably fine to offload.

Ask the question that unlocks real learning: "What did you think before you asked the AI?" If the honest answer is "nothing," the AI arrived too early.

This is also why the design of a learning tool matters so much. A chatbot optimized to answer questions as fast as possible will offload thinking by default. A tutor designed to guide rather than tell, breaking topics into chunks and asking students to work through each step, does the opposite. This is the philosophy behind LEAI, which uses context-aware conversation to help students discover answers rather than hand them over. If you want to see what guided AI learning looks like in practice, you can try LEAI free or explore the features.

The Takeaway

The cognitive offloading problem is real, but it is not a verdict on AI. It is a verdict on uncritical AI use. Students who treat AI as a thinking partner tend to think better. Students who treat it as a vending machine for answers slowly stop thinking at all. The direction of travel is set at home, in the classroom, and by the design of the tools themselves.

If you want to go deeper on the science behind learning that actually sticks, our guide to desirable difficulties explains why struggle builds skill, and our piece on fact-checking AI answers turns the research above into a practical habit students can start using tomorrow.

Sources

1. Gerlich, M. (2025). AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. SSRN.
2. Psychology Today (2025). AI Weakens Critical Thinking, and How to Rebuild It.
3. PsyPost (2025). AI tools may weaken critical thinking skills by encouraging cognitive offloading, study suggests.
4. Singularity Hub (2026). Is AI Making Us Dumber? A Review of the Evidence.