Critical Thinking and AI: Keeping Your Edge
Critical thinking and AI use turn out to be connected in a way that's worth taking seriously, not panicking about.
Critical thinking and AI use turn out to be connected in a way that's worth taking seriously, not panicking about. A 2025 study of 666 people found that heavier AI use tracked with lower scores on a standard critical-thinking test, and the connection ran through a specific mechanism: people were handing over the reasoning itself, not just the busywork.
That's a useful, precise finding, and it's different from the vague fear that "AI is making us dumber." Nobody in that study got dumber by using a calculator to add numbers. The concern is narrower and more specific: certain kinds of AI use quietly replace the mental habit of working through a problem yourself, and that habit is what critical thinking actually is.
What the research on critical thinking and AI actually found
The study, led by Michael Gerlich of SBS Swiss Business School and published in the journal Societies in 2025, surveyed and interviewed 666 people across a range of ages and education levels. Participants took the Halpern Critical Thinking Assessment, a standard psychometric test, and reported their AI usage patterns. (Gerlich, 2025, Societies)
The headline result: heavier AI users scored lower on the critical-thinking assessment, and the effect was strongest for tasks that require reasoning or reflection rather than simple recall. Cognitive offloading, the tendency to hand a mental task to an outside tool, was the mechanism linking the two. Younger participants used AI the most and scored the lowest on independent thinking. One finding worth holding onto: higher education attainment softened the effect, suggesting that people with more practice reflecting on problems held onto that skill better even with heavy AI use.
Gerlich is careful, and so should we be, that this is a correlational study. It doesn't prove AI use causes weaker critical thinking in a strict cause-and-effect sense. It's possible that people who already relied less on their own reasoning gravitated toward heavier AI use in the first place. But the size and consistency of the pattern, especially the offloading mechanism sitting right in the middle of it, makes this more than a coincidence to wave off.
Why this fits older research too
This isn't a new mechanism the AI era invented from scratch. Before AI, offloading research had already shown a version of the same pattern with search engines. In a 2011 study in Science, Betsy Sparrow, Jenny Liu, and Daniel Wegner found that when people expect information will be available online later, they remember the information itself less well, and remember where to find it better instead. The internet had become what researchers call transactive memory: a place to store facts so your own head doesn't have to. (Sparrow, Liu & Wegner, 2011)
AI extends that same trade into a different layer. A search engine offloads storage. AI can offload the analysis, the argument, the judgment call. That's a bigger ask, and it maps onto exactly the tasks the Gerlich study found were most affected: reasoning and reflection, not recall.
We go deeper on the mechanics of offloading itself, including where it helps and where it costs you, in cognitive offloading.
What "critical thinking" actually means here
It's worth being precise about what's at stake, because "critical thinking" gets used loosely. In the psychology sense it means things like: evaluating whether a claim is well supported, spotting a weak argument, weighing competing evidence, and reasoning through a problem you haven't seen before rather than pattern-matching to something familiar.
Those are exactly the skills that atrophy through disuse, the same way any skill does. If you hand the reasoning to AI often enough, the parts of the process that would normally sharpen through practice (generating your own hypotheses, noticing gaps in an argument, sitting with ambiguity before resolving it) simply don't get exercised as often. The tool isn't taking anything from you directly. You're just not doing the rep.
Where the risk actually shows up
Not every use of AI touches critical thinking. Asking AI to summarize a document you'll still evaluate yourself is low-risk. Asking it to draft an email you'll edit is low-risk. The higher-risk pattern is asking AI to decide: to weigh the tradeoffs, form the conclusion, or settle the judgment call, and then accepting that output without doing your own pass first.
This shows up at work more than people notice. If you'd like a fuller list of the specific habits that quietly hand over judgment on the job, we've catalogued them in signs you're relying on AI too much at work. A short version here: the tell isn't how often you use AI, it's whether you could still reach the same conclusion without it, and whether you'd notice if the AI's conclusion was subtly wrong.
How to protect your critical thinking while still using AI
You don't need to quit AI to keep this skill. The Gerlich study's own silver lining, that education and practiced reflection buffered the effect, points to the fix: keep exercising the reasoning muscle on purpose.
Form your own view before you ask. Read the question, sketch your own answer or argument, even a rough one, before you open a chat window. This keeps the reasoning happening in your head first, so AI is checking your thinking rather than replacing it.
Argue with the output. Don't just read an AI answer and move on. Ask what it left out, whether the logic actually holds, whether you'd reach the same conclusion by a different path. That's the habit of evaluation, and it's the one the research says is fading.
Pick a few things to always reason through yourself. Not everything needs protecting, but choose two or three recurring problems, a specific kind of decision at work, a category of argument you care about getting right, and commit to working through those without AI's help, at least on the first pass.
Notice when you're offloading judgment versus busywork. Summarizing, formatting, and drafting are lower stakes. Deciding what's true, what matters, or what to do next is the part worth keeping. For a broader playbook on this exact distinction, see how to use AI without losing your skills.
Curious where your own habits land? The Is AI Changing How You Think? assessment gives you an honest, anonymous read in a few minutes, grounded in the same research described here.
Five days, one change per day
The fix for this is rarely quitting AI outright, it's changing one habit at a time. The free 5-Day AI Reset walks it in five short emails: measure where AI shows up, take one task back, name the urge, sort helpful use from skill-replacing use, and set rules that survive a bad week. No quitting AI, no guilt.
Frequently asked questions
Does using AI actually lower critical thinking scores?
A 2025 study of 666 people in the journal Societies found heavier AI use correlated with lower scores on a standard critical-thinking test, with cognitive offloading as the mechanism. It's a correlational finding rather than proof that AI use directly causes the drop, and the pattern was consistent and strongest for reasoning-heavy tasks.
Is all AI use bad for critical thinking?
No. The concern is specifically offloading the reasoning and judgment, not using AI for storage, formatting, or busywork. Summarizing a document you'll still evaluate yourself is a different trade than asking AI to form your conclusion for you.
Who is most affected?
The 2025 study found younger participants used AI most and scored lowest on independent thinking, while higher education attainment softened the effect, suggesting people with more practice in reflective reasoning held onto the skill better.
How is this different from the classic worry about calculators or GPS?
Calculators and GPS offload narrower skills, arithmetic and navigation. AI can offload reasoning and judgment itself, which is a bigger trade and touches more of what critical thinking actually is. The mechanism, use it or lose it, is the same one researchers have studied for over a decade.
Can I use AI a lot and still keep my critical thinking sharp?
Yes, if you keep forming your own view before asking, argue with AI's output instead of accepting it, and protect a few recurring decisions as always-reasoned-through-yourself. The skill fades from disuse, not from AI's mere existence.
If you want an honest look at where your own AI habits sit, not a lecture, just a real read, the Is AI Changing How You Think? assessment is free, anonymous, and takes about four minutes.
Published August 2026.