Generative AI and critical thinking can come together in classroom activities that require students to examine, correct, and explain AI-generated material.
Dalia Alsaiid Abdelbaki’s (2026) study describes one such approach: students used AI to prepare lessons for their classmates, then discussed the outputs and the responsibilities involved in using them. The study offers a teaching idea worth considering, although its survey findings cannot establish that students became better critical thinkers.
I am interested in the responsibility this kind of task places on students. Preparing material with AI is relatively easy. Deciding whether that material deserves to be taught requires subject knowledge and judgement. A student who presents an explanation to classmates should be able to account for its accuracy, its meaning, and the choices behind it.
Generative AI and critical thinking need deliberate task design
An assignment can invite students to create something with AI while leaving their understanding largely invisible. A coherent presentation may contain accurate information, yet give a teacher little evidence of how the student evaluated it. Even a requirement to disclose AI use tells us little about the quality of those decisions.
The paper addresses this concern through an instructional approach the author calls co-creation. Students are expected to participate actively in developing the material, including examining AI outputs and correcting errors. Co-creation here describes the intended relationship between students, AI, and the learning task. It was not independently measured as an outcome.
This belongs within the wider discussion of critical AI literacy: learners need opportunities to question what AI produces and consider the consequences of using it. Calling students co-creators is a useful starting point, but the task still needs to show us where their judgement enters the work.
Summary of the Study
Alsaiid Abdelbaki (2026) examined students’ perceived AI literacy, learning benefits, and ethical awareness following an AI-integrated classroom activity at one university. The activity was implemented across two academic semesters.
Students used tools including ChatGPT, Google Gemini, and Microsoft Copilot to prepare and deliver short lessons to peers, followed by guided discussion about AI outputs and responsible use. Of the students invited to complete a questionnaire, 104 responded voluntarily.
Students generally reported positive experiences, including easier and more enjoyable learning, time savings, and improved work quality. Higher self-reported AI literacy was associated with greater perceived learning benefits and ethical awareness.
The analysis also identified an indirect statistical relationship linking AI literacy to ethical awareness through perceived benefits. Because these measures were collected at the same point after the activity, that relationship does not establish a sequence in which literacy produced benefits and benefits subsequently produced ethical awareness.
The findings describe students’ perceptions. Critical thinking was not directly assessed, and there was no control group or baseline comparison to establish improvement attributable to the activity. The single-institution, voluntary sample also limits generalisation.
I would therefore read the paper as an account of a structured teaching approach accompanied by positive student reports. Its title makes critical thinking central, but the evidence leaves the actual development of that ability open for further investigation.
Practical insights: make students’ decisions part of the work
The researcher recommends guided AI activities and reflective discussion. My suggested adaptation would be to make the checking process an explicit part of the lesson students prepare.
Alongside their teaching material, students could submit a brief record of a claim they verified, an explanation they corrected, or a suggestion they rejected. These additions are teaching proposals; the study did not test their effectiveness.
For example, a student preparing a short lesson could identify one AI-generated claim and check it against a course reading. If the claim is accurate, the student explains what supports it. If it needs qualification, the student revises it and explains why the original wording was misleading. The teacher can then examine the reasoning behind the final explanation.
That connects with evaluative judgement in the AI age. Students need criteria for deciding whether an output is useful and trustworthy. “Check the answer” is a broad instruction. Asking which claim needs evidence, whether an example fits the concept, or what the explanation leaves out gives students a more concrete task.
Peer teaching also creates an opportunity to ask follow-up questions. After a short lesson, classmates or the teacher could request clarification of a concept or ask why a particular example was selected.
Such exchanges may help reveal how well the student understands the material. They should be planned around the learning outcome and accompanied by appropriate supports for participation.
Ethical reflection can remain close to these decisions. Students might discuss how they credited assistance or what could happen if classmates relied on an unchecked explanation.
This gives responsibility a specific context. The next research step would be to examine students’ actual revisions, explanations, and checking practices, alongside their perceptions of the experience.
Reference
Alsaiid Abdelbaki, D. (2026). Generative AI as a catalyst for critical thinking: Reframing students from consumers to co-creators. Frontiers in Education, 11, Article 1952725. https://doi.org/10.3389/feduc.2026.1952725




