AI literacy for pupils: what they should learn

Pupils already use AI — for homework, presentations, translations, images. Often without knowing how the systems work, where their limits are and what happens to their inputs. School has the task of turning this use into a conscious, critical and productive competence — just as it does with reading, arithmetic and media literacy.

In Germany, the Standing Conference of Education Ministers described the competences pupils should acquire in a digital world in its strategy “Education in the Digital World” and has published recommendations on dealing with AI in schools. At European level, the European Commission's Digital Competence Framework describes similar areas.

AI literacy can be divided into five areas, covered in more or less depth depending on age.

Perhaps the most important skill is checking. Language models write fluently and confidently — even when they are wrong. Pupils should experience an AI inventing a source, mixing up a date or getting a calculation wrong. That works best through their own experience: ask a question whose answer you know and check the result.

Simple rules follow: always check important information against a second source. Ask for references and open them yourself. Be particularly sceptical with calculations and facts. And: do not adopt unchecked what you cannot judge yourself.

Many teachers worry that pupils hand over their thinking to AI. That worry is justified when AI is only used to produce finished solutions. But it can also be used in ways that promote thinking: as a conversation partner who asks questions; as a critic who finds weaknesses in your own text; as a practice partner who sets tasks.

Lesson tasks can be designed around this: document your own reasoning, improve an AI answer and justify why, or formulate a question that an AI can answer well. That makes working with AI a learning subject in its own right.

In primary school the focus is on first concepts: machines can recognise patterns but do not understand like people. Playful experiments — such as a program that recognises drawings — make that tangible. At secondary level, how language models work, critical checking and data protection are added. In upper secondary and vocational school, social, legal and professional questions become important, such as the rules of the EU AI Act or the use of AI in future jobs.

It is important that the use of AI services in class complies with the school's data protection rules. Many freely available services have age limits and process data outside the EU.

A platform can foster AI literacy by setting a good example in its own use of AI: labelling AI answers, binding them to sources, admitting errors openly and encouraging checking. On the Universe platform, presenter and narration are labelled as AI-generated; a fact check has two language models verify statements independently against archived sources and openly flags unverified statements. For teachers and staff, People Development includes the template “AI literacy in the company”.

For schools, Universe Classroom is planned as an AI assistant with its own models, curriculum alignment and teacher control; details are still being clarified.