Primary pupils need AI literacy too

The goal at primary level does not need to be abstract. It is enough for pupils to know that AI gets things wrong, to learn a simple way to check, and to understand when to ask a person rather than the AI. These three things match the three layers of literacy: being able to use it, being able to judge it, and being able to take responsibility.

Set the goal down at the level of concrete behaviour, and the teacher has something to teach and to observe. Saying 'develop critical thinking' in vague terms often gives a primary classroom no way in; change it to 'work out why this picture is fake' and the pupils have something to do right away.

Principles for the right age

For a primary AI literacy activity, success mostly comes down to whether it is pitched at the right age. Three principles can help you sift:

  • Concrete: use material pupils know, such as the school's name, Hong Kong landmarks or cartoon characters, so that the abstract idea that 'AI gets things wrong' becomes an example they can see.
  • Visual: junior primary pupils judge images faster than text. Starting with 'spot which one the AI drew' works better than explaining how a model works.
  • Clearly right or wrong: at primary level, choose activities with a clear answer, and avoid opening with an open-ended ethics debate, which suits senior primary through secondary better.

The difference between junior and senior primary

Within primary alone, junior and senior differ a great deal in how you go about it. In junior primary (Primary 1 to 3), it is best for the teacher to do the operating: the teacher puts the question to the AI and leads the whole class in judging whether the answer can be trusted, and pupils do not need to log in themselves. This avoids the privacy risk and turns the judging into a shared exercise for the whole class.

Senior primary (Primary 4 to 6) can begin some limited independent use, but it should take place inside an education tool with a preset framework, not directly on an open-ended chatbot. The tool sets the boundaries for the teacher, pupils operate, check and create within a safe range, and the teacher offers guidance from the side.

Two activities you can use straight away

The two activities below suit senior primary; junior primary can do a teacher-demonstrated version:

  • Spot the AI-generated image: the teacher mixes real photos with AI-generated images, groups of pupils find the giveaways (an extra finger, warped text, lighting that makes no sense), and then work out the flaws AI images commonly have, practising observation and media literacy at the same time.
  • Check a familiar question: have the AI answer a question the pupils already know the answer to, such as facts about their own school or local general knowledge, and the whole class compares the gap between the AI's answer and the truth, discussing why the difference is there and how you could check. Pupils see for themselves the AI getting it wrong with a straight face.

How this counts as a student AI literacy activity under the programme

The AI for Empowering Learning and Teaching Funding Programme asks every school to hold at least two activities that build students' AI literacy within the programme period, and this applies to primary schools too. For the activities above to become evidence in the report, set up the records in advance: which classes took part and how many pupils, samples of what pupils produced, and a simple three-to-five-question test before and after, to see whether ideas like 'the AI's answer must be correct' have changed.

The advantage at primary level is that you can fit a literacy activity into a regular General Studies, Chinese or Visual Arts lesson, run it with the whole class, and reach further than a sign-up competition outside class hours. Of the two activities, run at least one with a whole year group or whole class, so the aim of reaching everyone is not lost.