AI in the classroom should begin with a teaching problem and clear safeguards, not a requirement that every child use a chatbot. Teachers can explore low-risk preparation tasks, verify the output and decide whether the result improves learning. Students need age-appropriate AI literacy, including the ability to question an answer, recognise uncertainty and explain their own work.
AI in the Classroom: Begin with Teacher-Controlled Use
A sensible first experiment is to draft a short set of practice questions on a topic you know well. Check every question and answer yourself. If the tool creates more checking work than it saves, change the task or stop using it for that purpose.
UNESCO’s guidance on generative AI in education stresses privacy, age-appropriate use and human oversight. It is guidance, not a substitute for your school’s policy or local requirements. Before involving students, obtain the approvals your institution requires and check the exact provider’s current terms.
Four Useful Preparation Tasks
- Alternative explanations: ask for two ways to explain a familiar concept, then correct errors and choose language that fits the class.
- Question drafts: request a small set of problems, an answer key and possible misconceptions. Solve the problems independently.
- Activity variations: adapt a task for different reading or access needs without changing the intended learning goal.
- Planning support: generate a checklist of materials or lesson transitions, then compare it with the real room, time and learners.
Do not upload identifiable student work, health information, safeguarding records or private family details to an unapproved service. Removing a name alone may not anonymise a distinctive story. Use synthetic examples when testing a workflow.
A Prompt That Keeps the Teacher in Control
Try: “Draft five questions about equivalent fractions for learners who can already identify halves and quarters. Include two visual-description questions, an answer key and one likely misconception for each. Do not claim curriculum alignment unless I provide the standard. Flag any assumptions.”
This is a starting instruction, not a guarantee of accuracy. Check that the numbers work, the wording is unambiguous and the visual descriptions could actually be drawn. Remove unnecessary cultural assumptions and make sure the task does not depend on resources the class lacks.
Save the verified version in your ordinary lesson materials. Keep a note of meaningful AI assistance where your school requires disclosure. Do not treat the model’s own confidence or self-check as independent validation.
Teach Verification Without Student Accounts
Prepare a short, teacher-selected example containing a plausible factual error. Tell learners that the task is to investigate an answer, not to guess whether a person or a machine wrote it. Provide reliable reference material and ask them to mark claims they can verify, claims they cannot verify and claims that are incorrect.
A simple example is an incorrect arithmetic explanation. Learners can solve the problem with counters or a drawing, compare the reasoning and write a corrected explanation. Keep the example suitable for the age group and do not expose children to harmful material for the sake of a demonstration.
Close by asking: “What evidence changed your mind?” This develops a habit that applies to textbooks, websites and conversations as well as AI output.
Set Task-Specific Rules
“Use AI responsibly” is too vague on its own. For each assignment, state what assistance is allowed, what must be done independently and how assistance should be acknowledged. A brainstorming task may permit suggestions while an assessment of unaided writing may not.
Ask students to retain meaningful drafts, notes or design decisions where appropriate. Disclosure should explain the tool’s role, not pretend an AI answer is a reliable source for factual claims. Students still need to consult and cite the underlying evidence required by the task.
Offer an equivalent route for learners who cannot or should not use the tool. Access to a paid service should not become an undeclared condition for doing well in class.
Assessment and Academic Integrity
Use a combination of classroom observation, short explanations, process work and appropriate independent tasks. These methods help you assess understanding regardless of whether AI tools exist. Avoid accusing a student solely because writing “sounds like AI” or a detector produces a score.
If a concern arises, follow the school’s assessment process, discuss the work with the learner and consider the available evidence fairly. Language background, disability and writing style should not be treated as proof of misconduct.
Do not delegate high-stakes grading, safeguarding judgements or decisions about a child’s needs to an unreviewed model. A generated suggestion is not professional accountability.
Run a Small, Measurable Pilot
Choose one class task and record preparation time, checking time, errors found and what students could explain afterwards. Compare it with your usual method. A faster worksheet is not a success if the questions are weaker or the teacher has less understanding of how they were produced.
Review privacy, access and learning outcomes before expanding. Keep an offline backup and make it easy to stop the trial. The goal is a better lesson, not maximum AI usage.
Find more planning support on our Teachers hub and age-appropriate explanations on our Artificial Intelligence hub.
