AI in the classroom: opportunities, limits, and practical rules

AI in the classroom is already here, even when a school has not discussed it openly yet. Students use it to summarize, translate, answer questions, prepare for oral tests, or revise writing. Some use it well, many use it without clear criteria, and others hide it because they do not know whether it is allowed.

The point is not choosing between a total ban and naive enthusiasm. The teacher's role is more concrete: explain what artificial intelligence can do, where it fails, which data must not be entered, when use must be declared, and how to verify what it produces.

UNESCO stresses this exact need: generative AI in education requires human oversight, responsibility, equity, data protection, and the development of critical competencies. UNESCO Guidance for generative AI in education and research

AI in the classroom: real opportunities, not vague promises

Artificial intelligence becomes useful when it enters a precise learning task. It is not enough to say "let's use AI" in general. It is better to say: "let's use it to turn these notes into a first map", "let's use it to generate review questions we will check", or "let's use it to compare two explanations and decide which one is clearer".

For a teacher, the most concrete opportunities are four:

1. Faster preparation: start from PDFs, slides, handouts, or notes and get a first structure.
2. Controlled personalization: create variants of the same material with a different load, vocabulary, or level of detail.
3. Accessibility: offer multiple channels, such as text, maps, audio, examples, images, and keywords.
4. Feedback and revision: help students improve a draft, identify unclear points, and check the quality of their reasoning.

CAST's UDL Guidelines invite educators to offer multiple means of representation, action, and engagement. AI can help create these formats, but the teacher still decides what the class actually needs. CAST UDL Guidelines

For a more material-preparation-focused workflow, read AI for teachers: creating inclusive materials without losing hours.

Where AI actually helps a teacher

A good use of AI almost always starts from real material: the handout you will use tomorrow, the slides you have already prepared, lesson notes, an audio recording, a text that needs careful simplification, or a map that needs to be made more readable.

Useful examples include:

1. turning a long explanation into a lesson outline;
2. extracting keywords from a chapter;
3. creating self-check questions to correct with the class;
4. proposing a first mind map to review;
5. generating different versions of an activity for recovery, review, or enrichment.

The value is not the first AI answer. The value is having a fast draft to which the teacher can apply professional judgment: remove, correct, change examples, check sources, and adapt vocabulary to what has already been used in class.

Limits and risks: what students need to understand

Students need to know that AI can produce convincing but wrong text. It can invent sources, oversimplify, miss the class context, use an unsuitable tone, or give an answer that sounds correct without proving any learning.

The main risks are five:

1. Errors and fabricated sources: every important claim needs verification.
2. Privacy: names, diagnoses, marks, family details, and sensitive information must not be entered into unauthorized tools.
3. Dependence: if AI does everything, students do not train retrieval, selection, argumentation, or revision.
4. Weaker critical thinking: a ready-made answer may look more solid than it is.
5. Opaque assessment: if AI use is not declared, the teacher cannot see what the student produced and what the tool produced.

The European framework for AI follows a risk-based approach and includes transparency and governance obligations for different uses of artificial intelligence. For schools, this does not mean turning every teacher into a lawyer. It means recognizing that tools, data, and responsibility are not neutral. European Commission AI Act overview

For personal data, GDPR remains the reference point: before entering information into a digital service, schools need to understand which data is being processed, why, and under which safeguards. European Commission GDPR rules

Simple classroom rule: if information concerns a real person, diagnosis, assessment, behavior, or family situation, do not enter it into an AI tool without authorization and without a clear basis defined by the school.

Practical rules to write before the activity

Rules work when they are few, readable, and linked to examples. A class does not need an abstract twenty-page policy. It needs to know what is allowed in that specific assignment.

A practical structure can be:

1. Declare use: write whether AI was used and for which step.
2. Do not submit the first answer: rework, verify, and explain choices.
3. Check sources: if AI mentions a fact, go back to the original source.
4. Protect data: do not enter names, diagnoses, marks, or personal details.
5. Keep the thinking: AI can help preparation, but the thesis, examples, and conclusion must remain yours.

These rules need to match the activity. In a writing task, you can allow AI for outline and revision, not for the final submission. In a research activity, you can use it to generate initial questions, but not as a source. In review, you can use it to build a map that students correct with the textbook open.

How to build a classroom agreement

A classroom agreement is not only about preventing copying. It turns AI into a digital citizenship topic. Students learn that a tool can be powerful and fallible, useful and risky, convenient and still governed by rules.

You can build it in 30 minutes:

1. Show two AI answers, one useful and one problematic.
2. Ask the class to identify mistakes, useful parts, and claims to verify.
3. Define three allowed uses and three non-allowed uses together.
4. Decide how students will declare AI use at the end of an assignment.
5. Update the agreement after the first real activity.

The declaration can be simple: "I used AI to generate an outline and three review questions. I checked the information in the textbook and rewrote the explanation in my own words." It is short, but it changes the quality of the work because it makes the process visible.

AI and maps: a good use because it forces reworking

One of the most sensible uses of AI in class is turning content into maps to be checked. A map is not a ready-made answer to copy: it shows hierarchies, connections, keywords, and gaps. It forces students to ask whether the order is correct, whether a step is missing, and whether an example actually fits.

With Kiuwo, the workflow is this: start from real material, generate a first mind map, then review it. The teacher can ask students to correct one branch, add a source, remove unnecessary details, turn the map into an oral explanation, or compare it with the textbook.

This is different from "give me the answer". It is a use that supports better thinking: the structure appears, but it needs to be understood, verified, and made one's own.

If you start from a ChatGPT conversation, you can also use the workflow described in Kiuwo inside ChatGPT. If the topic is inclusion, connect this work to strategies for teaching students with SLD.

Checklist before using AI in a lesson

Before proposing an AI activity, check these points:

1. What is the learning objective, independent of the tool?
2. Which part of the work can AI support?
3. Which part remains the student's responsibility?
4. Which data must not be entered?
5. How will AI use be declared?
6. Which source will be used to verify the content?
7. How will you assess process, revision, and final product?

If you cannot answer one of these questions, the activity is not ready yet. Make it narrower: one prompt, one source, one map, and a final discussion about what AI did well and what it got wrong.

Frequently asked questions about AI in the classroom

Sources used

• UNESCO Guidance for generative AI in education and research
• UNESCO Artificial intelligence in education
• European Commission AI Act overview
• European Commission GDPR rules
• CAST UDL Guidelines
• Unsplash License

Want students to rework ideas, not copy answers? With Kiuwo you can turn texts, notes, and lessons into mind maps to check, correct, and use as a base for review, discussion, and assessment.

Try it now, it's free