AI literacy starts before the prompt
Artificial intelligence is becoming part of everyday educational conversations. Teachers may encounter tools that generate text, create images, translate languages, summarise information, provide feedback or adapt activities. Some can save time and open creative possibilities. Others may produce inaccurate, biased or unsuitable results.
The important professional skill is not simply knowing how to operate an AI tool. It is knowing when to use it, when not to use it and how to question its output. The European Commission’s ethical guidelines for educators emphasise critical, ethical and responsible AI literacy in education.
Before introducing any classroom AI tool, ask five questions.
1. What learning problem is the tool solving?
Begin with the learning objective, not the technology. A tool may generate a quiz in seconds, but speed alone does not make the activity educationally valuable. Ask what learners should understand, practise or create, and whether the tool genuinely supports that aim.
An AI tool might help learners compare different explanations of a scientific concept, support language learners with alternative sentence structures or provide a starting point for a creative project. In each case, the teacher remains responsible for designing the task and deciding what successful learning looks like.
- What will learners know or be able to do after the activity?
- Does the tool support thinking, practice or communication?
- Could the same objective be achieved more clearly without AI?
If the answer to the final question is yes, that is not a problem. Responsible digital education includes choosing the simplest effective method.
2. What does the tool actually do?
“AI-powered” is a broad description. A tool may generate content, classify responses, recommend resources or analyse patterns. Each function creates different opportunities and risks. Test the tool with several examples. Does it explain its limitations? Does it produce different answers to similar questions? Does it show where information comes from? Can users correct it or challenge its response?
Explain to learners that generated content is not automatically true, original or appropriate. AI systems can produce fluent answers that contain errors. Asking learners to verify, compare and improve outputs can turn this limitation into a valuable critical-thinking activity.
3. What personal data might be involved?
Never assume that a classroom tool is suitable for student information simply because it is easy to access. Before using it, check what data the service collects, why it collects it, how long it keeps it and whether content is used to improve the system.
Use general, fictional or anonymised examples whenever possible. Avoid entering names, photographs, individual support needs, assessment details or identifiable student work unless the school has completed the required checks and has a clear lawful basis for doing so.
The Commission’s AI education guidelines connect responsible AI use with data protection, transparency and the rights of learners. School policy and national requirements should always be consulted before using a tool with personal data.
4. Could the output disadvantage someone?
AI systems learn from data and patterns that may reflect existing inequalities. An automated suggestion may misunderstand a learner’s language variety, cultural background, disability, communication style or level of access to technology. A generated image may reinforce stereotypes. A feedback system may reward a narrow form of writing.
Ask whether the task gives learners a fair opportunity to participate. Provide a non-AI alternative where necessary, especially when access to accounts, devices or paid features is uneven. Review examples for stereotypes, exclusionary language and assumptions about family life, culture or ability.
Teachers should be cautious about using AI to make high-impact judgements about learners. A generated recommendation can inform professional thinking, but it should not replace human judgement, dialogue and evidence.
5. How will learners show their own thinking?
The best classroom use of AI leaves room for learner agency. If a tool produces the entire answer, learners may complete the task without understanding the process. Design activities that require explanation, decision-making, revision or reflection.
Students might submit an initial question, the AI response, a fact-checking record and a final version with a short explanation of what they changed. They might compare two generated answers and identify which is better supported. They might use AI to suggest possibilities, then select and justify their own solution.
This approach makes the learning visible and helps establish clear expectations about acceptable and unacceptable uses of AI in assignments.
A short pre-use checklist
Before using a new tool, write down five brief answers:
- The learning objective is…
- The tool contributes by…
- The main accuracy or bias risk is…
- The data we will and will not enter is…
- Learners will demonstrate their own thinking by…
If these answers are unclear, delay the activity and test the tool further. AI literacy is not about adopting every new service. It is about making informed choices that protect learners and improve learning.
