If your child has ever finished a writing assignment in seconds by handing it to a chatbot, you have felt the small unease underneath the convenience. This week a study out of Italy put a shape to it. Researchers found that generative AI shifts a student’s mental work away from producing an answer and toward judging one, and that under mental fatigue, even strong critical thinking stops guarding against passively accepting whatever the tool produces. The study looked at university students, not children, and its authors are careful about how far it reaches. But it lands on a question every parent of a school-age child now faces: when a tool makes the thinking easier, what happens to the learning the thinking was supposed to build?
A new study found that when people are mentally tired, even strong critical thinking stops protecting them from swallowing whatever AI hands over. The finding points at something bigger: when a tool removes the effort, it often removes the learning the effort would have built.
The rush of AI into schoolwork raises the questions parents are already asking at the kitchen table.
Common questions
Is it bad for my child to use AI for homework?
How do I tell if an AI tool is helping or hurting my child’s learning?
My child is exhausted after school and reaches for AI. Should I worry?
How do I find out where my child’s learning actually needs support?
New study: when a student is tired, even sharp critical thinking stops guarding against accepting whatever AI hands over. When a tool removes the effort, it often removes the learning too.
What happened
The study, reported this week by the news outlet Devdiscourse, was published in the European Journal of Investigation in Health, Psychology and Education by Marco Zuin of the Istituto Universitario Salesiano Venezia and Vanessa Donadel of LUISS University. The researchers surveyed 383 Italian university students, most of them around age 22, and found that 332 of them, nearly nine in ten, had used a tool such as ChatGPT for study in the previous three months. The question they set out to answer was not whether students use AI, but how readily they accept what it hands back without changing a word.
Their finding has two halves. Critical thinking, as they put it, generally reduced passive acceptance of AI output. But that protective value weakened under conditions of decision fatigue, the worn-down state that sets in after a long run of choices. Put plainly, when a capable, critical student is mentally tired, the guardrail that keeps them from swallowing AI output whole stops holding.
The researchers are careful about what this shows, and a parent should be too. The study is a one-time snapshot of a self-selected group in a single country, and the authors state that it does not establish causation. It explains only a small share of why some students accept AI output more passively than others. It is a real, peer-reviewed finding and a modest one, and it is a starting point for a question rather than the answer to it.
Even in the presence of high levels of critical thinking, when individuals experience decision fatigue, the protective function of critical thinking against the passive use of AI is substantially weakened.
Laura Lurns · Learning Success expert
The frame the science supports
Here is where the popular story goes wrong, and it is worth being precise about whose story it is. The outlet reporting this study did not celebrate AI for making thinking effortless; it led with the concern. The celebration lives somewhere else, in the back-to-school marketing and the study-smarter-not-harder pitches that treat friction as the enemy and a frictionless path to the finished answer as the goal. That assumption is the thing the learning science quietly contradicts.
Decades of research on how people actually learn point one direction, though it is worth saying that none of it tested a chatbot, so treat this as the mechanism rather than a measured verdict on AI. In a 2006 study, Roediger and Karpicke found that students who effortfully pulled information back out of memory held onto it far longer than students who simply re-read the same material, even though re-reading felt easier and more productive in the moment. The effort of retrieval was not in the way of the learning. It was the learning. In long lines of research on tutoring software, running years before chatbots arrived, VanLehn in 2011 and Kulik and Fletcher in 2016 found the same split: programs that engaged a student’s reasoning one step at a time helped nearly as much as a human tutor, while programs that only checked whether the final answer was right barely did better than no help at all.
Underneath sits an old and stubborn principle: practice carries over to a real task only to the degree it is built from the same mental steps that task demands. Practice reading a finished, fluent answer builds the skill of reading finished answers. It does not build the skill of working one out. This is also why sticking with difficulty is not a character virtue bolted onto learning but part of the machinery of it. And none of this makes AI the villain, which is the whole point. A tool built to ask your child a question back, to make them attempt the step before it responds, to show one worked example and then stop, is doing the opposite job. What matters is not whether a tool runs on AI. It is whether it makes the child do the thinking or does the thinking for them.
Key takeaways
- The ease is the risk: removing the effort of producing an answer often removes the learning the effort would have built.
- Tired minds drop their guard: fatigue weakened even strong critical thinking against accepting AI output whole.
- The one test that travels: does the tool make your child think, or think for them?
What it means for your child
Now the bridge the study itself does not make. Its students were adults. But your school-age child is walking into the same design, and the numbers on that are moving fast. A RAND survey of American youth, which includes students from the middle-school grades up, found that using AI for homework climbed sharply through 2025, and that a majority of students themselves agreed that leaning harder on AI could harm their own critical thinking. Children sense the trade even when they take it.
The practical move is smaller than understanding how a large language model works, and you do not need to become an AI expert to make it. It is one question you carry to any tool your child uses: when the answer appears, who did the thinking? If the finished answer shows up before your child has tried, the tool has taken the part that would have built the skill. If it makes them attempt the problem first, offers a nudge, and only then responds, it is working with them rather than for them. Ease is not the enemy, and a tired child handed a hard night of homework is not failing a character test. The enemy is ease in the one place where the effort was the point.
This is the ground Learning Success is built on: children learn most from effort they see through, not from work they are handed. The approach starts by asking what you are seeing at home and builds a plan around the skills your child needs to practice, with you coaching that practice rather than a screen supplying the answers. If the question this raises is what you would actually do, here is how that works. AI is arriving in classrooms and on phones faster than anyone is checking it, and your child will meet many of these tools. You do not have to vet the algorithm. You have to watch one thing: whether the tool is making your child think, or thinking for them. That question travels to every app, in every subject, no training required.
Your child is capable of hard thinking, and the effort a hard problem demands is where the learning happens, not an obstacle in front of it. The villain here is not AI, and it is not a tired child reaching for an easier path at the end of a long day; it is an old and comfortable idea in a bright new interface, that the smoother the road to the answer, the better the learning must be. Learning Success was built on the opposite conviction, that effort seen through is what builds a skill, and that you are the person best placed to protect it. Our All-Access membership opens an assessment that asks what you are seeing at home and returns a roadmap naming what to build first.
See what All Access gives your child
See MTSS for schools
Is your child struggling in school?
Get your free personalized learning roadmap
You describe what you see at home. We turn it into a plan you start this week.
- Answer 5 short questionnaires about what you already notice, 30–45 minutes at your own kitchen table
- Your child sits no test and gets no score: nothing to schedule, nothing for them to dread
- You do the answering, the AI does the writing, and a person reviews it before it reaches you
- Access all 40+ courses instantly: reading, math, focus, processing and more, with new ones added regularly
Why we use AI, plainly: it writes from a knowledge base our team maintains and audits. We work through it line by line and pull anything the evidence stops supporting. The roadmap you get on Tuesday reflects what we corrected on Monday, and a human still reads it before you do.
Your school district must evaluate your child free of charge if you ask in writing, whatever your income and whatever the outcome (US, 34 CFR 300.111 and 300.301(b)). That route takes time and answers a different question than you do. This one starts today, from what you already know.
Your answers stay yours. We do not sell your personal information, and we do not hand identifiable assessment data to outside AI companies to train their models.
A screener is a starting point, not a diagnosis. If your child might need formal accommodations (an IEP or 504 plan), or you suspect a vision, hearing or medical cause, pursue a professional evaluation too. That is the only route to those supports.
References
- Zuin, M. & Donadel, V. (2026), European Journal of Investigation in Health, Psychology and Education — From Critical Thinking to Passive Acceptance: The Moderating Role of Decision Fatigue in Students’ Use of Generative AI
- Devdiscourse — The New Education Challenge: Staying Critical When AI Makes Thinking Easier
- RAND Corporation — American Youth Panel (report RRA4742-1): youth use of and attitudes toward AI
