The real risk: outsourcing the thinking
Critical thinking grows through effort: weighing evidence, noticing gaps, changing your mind. If a chatbot hands over a finished answer and a student copies it, that effort never happens. Two habits cause most of the trouble:
- Accepting an answer because it sounds confident.
- Skipping the productive struggle of working out an idea.
AI tools can state false information confidently, which makes the first habit especially risky. See Why does AI make things up?
When AI can support thinking
The same tool can be a sparring partner. Students think harder when they have something to push against: an AI answer to question, compare with their own, check against sources, or argue with. Here the AI output is material to examine rather than a finished product. The difference between a helper that explains and one that simply answers is explored in AI as a tutor vs. AI as an answer machine.
Classroom moves that keep thinking in the student's hands
- Predict first. Before asking AI, students write what they expect the answer to be and why.
- Explain in your own words. After reading an AI answer, students restate it without looking and say what they are unsure about.
- Find the error. Give students an AI response, sometimes with a planted mistake, and ask them to find and fix problems using real sources.
- Argue against it. Students write the strongest counterargument to the AI answer.
- Compare. Students ask two tools, or AI and a textbook, and explain differences.
- Show the trail. Students record prompts and what they kept or rejected.
Where the WISE pillars apply
Wield It asks students to decide what to hand to AI and what to keep for themselves, before writing a prompt. Scrutinize It builds the habit of checking answers against real, non-AI sources. Endorse It makes students accountable for the final work. Together they turn AI use into a thinking exercise. See What is the WISE framework? and What is AI literacy?
Treat it as a design choice
The question is less whether AI hurts thinking and more whether a given task makes thinking necessary. A task that rewards a polished answer invites outsourcing. A task that rewards prediction, checking, explanation and critique invites thinking. Students also need to hear the reasoning behind the design, so explain why you are asking them to predict, check and critique: the point is to keep their own thinking in the work. Start with one lesson: add a prediction step before AI is used and an error-finding step after. Observe what students notice, and adjust from there.
Quick answers
Should AI be kept away from students to protect critical thinking?
Not necessarily. Thinking is protected by task design: ask students to predict, verify, explain and critique rather than just receive answers.
What is a quick activity to try?
Give students an AI answer containing a planted error and ask them to find it and correct it using a real source.