Practical teaching ideas

AI Literacy for Teachers: 7 Classroom Activities to Get Started

How can students decide whether an AI answer is useful, accurate or fair? These seven classroom activities give them opportunities to investigate, compare and explain their choices. You can start with printed examples, a familiar lesson topic and twenty minutes of discussion.

Four educators reviewing a laptop together during a digital planning workshop
Learn through questions Discussing an answer and checking its evidence are useful starting points for AI literacy.

What is AI literacy for teachers and students?

AI literacy means developing enough understanding to recognise AI uses, examine their limitations and make informed decisions about them. In a classroom, that includes checking a claim, questioning an assumption, deciding what information to share and explaining how a tool contributed to a piece of work.

UNESCO’s AI competency framework for students connects technical understanding with ethics, human agency and critical judgement. The European Commission’s guidance for educators also offers questions and scenarios for making responsible decisions about AI and data in schools.

The activities below are suggested starting points, rather than a complete AI curriculum. Most focus on generative AI: systems that produce content such as text or images. Students do not need to learn coding before they can begin examining that content critically.

Before you begin: use examples you have reviewed and fictional details throughout. All seven activities can run without student accounts, using printed materials or a teacher demonstration. For any live use, follow your school’s approved-tool policy and the service’s age and access requirements. Keep identifiable student work and personal information out of demonstration prompts. Label teacher-written simulations honestly; do not present them as genuine AI outputs.

1. AI detectives: what do we actually know?

Time: 15–20 minutes. Materials: scenario cards and three category labels.

Learning goal: distinguish evidence about a system from assumptions about its appearance.

Prepare cards describing a basic calculator, a fixed alarm, a system trained on labelled images to recognise objects, a chatbot generating a story and an app that recommends videos without explaining how.

  1. Label three spaces: “Evidence of AI”, “Fixed rules or calculation” and “Need more information”.
  2. Ask pairs to sort the cards and underline the words that support each decision.
  3. Have pairs write one question they would ask about an uncertain example.
  4. Discuss which extra information would make them move a card.

Teacher note: for these examples, specify that the calculator and alarm use fixed instructions. The trained image classifier and generative chatbot belong in the AI category. The recommendation app needs more information: its function alone does not reveal how it works. Real products can combine several kinds of technology.

Reflection: “What evidence would change your mind?”

Adaptation: use picture cards with younger learners; ask older students to examine a product’s explanation of one feature.

2. Check the claim, not the confidence

Time: 25–30 minutes. Materials: a short sample answer, a textbook extract and a second reliable reference.

Learning goal: evaluate individual claims using evidence.

Choose a topic students have already studied. Prepare a short AI-generated response and check it yourself first. If it contains no useful points for debate, write a clearly labelled practice passage with accurate, inaccurate and unsupported statements. Prepare an answer key with sources.

  1. Students underline three factual claims in the passage.
  2. Pairs mark each claim “supported”, “contradicted” or “not yet verified” using the reference materials.
  3. They record where they found the evidence and whether it supports the exact wording.
  4. They rewrite one sentence to correct an error or make uncertainty clear.

A polished explanation can still contain mistakes. If a response names a source, students should check that it exists and supports the claim. Asking the same chatbot whether it is correct is not independent verification.

Reflection: “Which claim sounded convincing before you checked it?”

Adaptation: give younger learners two short reference cards; ask older students to find and assess an additional source.

3. Run a prompt comparison lab

Time: 20–25 minutes. Materials: two prompts, their prepared responses and a topic reference sheet.

Learning goal: judge how instructions affect usefulness without treating clearer output as proof of accuracy.

Compare these two prompts:

  • “Explain the water cycle.”
  • “Explain evaporation, condensation and precipitation to an 11-year-old in 120 words. Include one everyday example and finish with a question that checks understanding.”
  1. Before showing the responses, agree on criteria: accurate content, understandable language, a useful example and a relevant question.
  2. Give students both responses and ask them to highlight evidence for each criterion.
  3. Ask pairs to change one part of a prompt and predict what that change might achieve.
  4. Compare a prepared follow-up response, or demonstrate it with an approved tool, then verify its subject content.

Treat this as a small comparison, not proof that one prompt always works best. Generated responses can vary, and extra instructions do not guarantee that every requirement will be followed.

Reflection: “What improved, and what still needs checking?”

Adaptation: change the topic to your current lesson and use shorter responses for less confident readers.

4. Find the missing perspectives

Time: 25 minutes. Materials: a sample school activity plan and fictional learner-context cards.

Learning goal: notice assumptions and omissions that affect who can participate.

Use a reviewed AI-generated plan, or this teacher-written simulation: “For homework, film a garden wildlife tour on your smartphone, edit it with a paid app and upload it tonight.”

  1. Students list what the task assumes about access to equipment, money, outdoor space and time.
  2. Give groups fictional contexts: a learner sharing one household device, someone living without a garden, or a pupil with limited internet access.
  3. Ask each group to identify a barrier and redesign the task while preserving its learning goal.
  4. Compare alternatives such as observing a school space together, drawing observations or offering an offline submission.

Explain that bias can appear through whose circumstances are treated as normal. An omission in one output is something to investigate; it does not establish how every output from that system will behave. Keep the discussion about the fictional scenarios rather than classmates’ circumstances.

Reflection: “Whose situation did this plan overlook?”

Adaptation: older learners can compare several plans and distinguish repeated patterns from isolated examples.

5. Become a privacy editor

Time: 15–20 minutes. Materials: fictional prompt cards and pens.

Learning goal: remove unnecessary personal information while keeping a task understandable.

Start with an invented prompt: “Write an invitation for Mila Perer, who lives at 14 Example Lane and attends Example School. The party is on Saturday at 3 pm. Her family’s phone number is [phone number].” Keep the entire exercise on paper.

  1. Students highlight identifying details and discuss whether an invitation-writing tool needs them.
  2. Pairs replace them with placeholders such as [name], [location] and [contact details].
  3. They produce a reusable invitation template that can be completed privately afterwards.
  4. Discuss why removing a name alone may leave other identifying clues.

Apply the same question to photographs, uploaded documents and copied messages: what information would the service receive? Removing details does not automatically make a tool suitable for school use; school policy still applies.

Reflection: “What is the minimum information needed for this task?”

Adaptation: younger learners can sort cards into “needed for the template” and “add privately later”.

6. Put AI feedback to the test

Time: 25–30 minutes. Materials: a teacher-written paragraph, a task rubric and prepared AI feedback.

Learning goal: accept, adapt or reject feedback with a reason.

Write a fictional paragraph for a familiar assignment. Obtain feedback through an approved tool before the lesson, or prepare a labelled simulation. Include suggestions students can evaluate against the actual task, such as adding evidence, replacing precise words or making the tone more formal.

  1. Students read the rubric and identify one strength and one weakness themselves.
  2. Reveal the feedback and ask them to mark each suggestion “accept”, “adapt” or “reject”.
  3. Pairs justify a decision with evidence from the paragraph or rubric.
  4. They revise one sentence and write a short explanation of their change.

Give credit for the explanation. A fluent rewrite is less informative about learning than a student’s ability to explain why a change helps the reader. Finish with a disclosure sentence: “We used AI feedback to consider ___; we changed ___ because ___.”

Reflection: “Which suggestion would make the writing less suitable for this task?”

Adaptation: use one sentence and two suggestions with younger pupils, or compare feedback on an argument with older students.

7. Create and test a class AI agreement

Time: 30 minutes. Materials: the school’s AI guidance, task scenarios and a shared writing space.

Learning goal: explain when AI use supports a learning objective and when independent work is required.

  1. Explain the school’s existing rules and one assignment’s learning goal.
  2. Give groups scenarios: brainstorming headings, requesting a hint, generating an entire response, or checking a claim against a textbook.
  3. Ask them to decide what is permitted for that assignment and justify their decisions.
  4. Draft four or five clear class expectations, then test them against a new scenario.

Useful expectations include checking whether AI is permitted for the task, protecting personal information, verifying factual claims, explaining any permitted assistance and being able to discuss the submitted work. Agree on how learners can ask for clarification when a rule is unclear.

Compare two assignments: an independent writing assessment and a lesson evaluating generated text. The appropriate role of AI can differ because the learning goals differ. Keep the agreement consistent with school policy.

Reflection: “Could someone follow our rule without knowing what we meant?”

Adaptation: younger classes can create illustrated expectations; older learners can add examples of transparent AI-use statements.

Check understanding with a three-minute exit ticket

After an activity, ask learners to complete these sentences:

  • “One thing I would check before using an AI answer is ___ because ___.”
  • “One decision that remains my responsibility is ___.”
  • “Evidence that would change my mind is ___.”

Look for specific explanations rather than repeated slogans. In the next lesson, offer a fresh example and see whether students can apply the same reasoning with less support.

Start with one activity in a subject you know

Choose a lesson where you can confidently evaluate the content. Prepare the examples, run one activity and note where students need more support. You can return to evidence, privacy and responsibility across different subjects instead of trying to cover everything in one session.

For related ideas, explore our digital citizenship classroom activities. Teachers wanting to develop these approaches further can compare our Critical Thinking with AI and Ethical AI in Education training programmes.

Develop your teaching practice

Explore AI literacy training for teachers

Understanding Academy’s AI Literacy for Teachers course explores accuracy, bias, privacy and responsible classroom use. Develop an adaptable activity and practical guidance for your educational setting.

Further reading for teachers

These resources provide broader guidance on AI education. The activities above are suggestions from Understanding Academy, not official UNESCO or European Commission lesson plans. Sources checked on 8 October 2026.