Unplugged AI Activities for Kids

Unplugged AI activities use movement, cards, discussion or role-play instead of a computer. They are useful when devices are unavailable or when a teacher wants students to reason about a concept before using software. AI4K12 organizes K–12 learning around ideas such as perception, representation, learning, interaction and societal impact; a paper activity can make one of these ideas easier to discuss without pretending to reproduce a full AI model. For a classification game, give teams labeled examples and ask them to invent a rule that separates categories. Introduce a new example that does not fit neatly, or remove examples from one group. Students can notice how the training set and chosen labels shape a decision. Rotate roles so children experience how different examples can change the rule. Explain that a classroom classifier is a simplified analogy: real systems use mathematical representations and training procedures that the game does not capture. A recommendation activity can use a branching paper path: a student selects a story card, then receives a suggested card based on a rule. Ask what the rule rewards, which options are missing and how a different goal changes the output. Avoid saying that the classroom path is how any one company’s algorithm works. Use it to introduce the idea that design goals and observed behavior influence recommendations. End with reflection. Have students identify evidence, limits and possible effects on people. Use made-up examples rather than collecting children’s personal images, voices or sensitive details. Check that rules are understandable, group roles are accessible and no student is labeled by a classroom exercise. An unplugged activity should prepare questions for deeper study, not replace lessons on real data, model evaluation or responsible use.