Where Elementary Students Are Starting From
Young students have a working theory of AI that is almost entirely magical. The voice assistant answers because it knows. The game character moves because it wants to. The recommendation system suggested that video because it understands. The mechanism is invisible, and invisible mechanisms become magic in the absence of any better explanation.
The goal at K-5 is not to replace magic with a complete technical account of machine learning. It is to plant the first seed of accurate intuition: things that seem to think are actually following rules. Those rules were written or learned by people. The machine is doing something, not knowing something.
That single shift — from "it knows" to "it follows rules" — is the most important conceptual move available at this level, and it is fully achievable with elementary students.
Decision Trees at the Elementary Level
Decision trees are the ideal entry point for AI instruction at K-5 because they are concrete, visual, and buildable by hand. Elementary students do not need to know the term "decision tree" — they can work with the structure before they have the name for it.
The Animal Sorting Game
Give students a collection of animal cards or pictures. Ask them to sort the animals into groups using only yes/no questions — but they have to use the same question at the same point for every animal. "Does it have four legs?" applies to every animal at the first branch. Students quickly discover that some questions are more useful than others: "Is it cute?" is hard to answer and does not produce clean groups, while "Does it have fur?" splits the set much more reliably.
This is the intuition behind good splits, experienced physically before it is named. The follow-up question — "How do you know which question is better?" — opens a productive discussion about what makes a question useful for sorting.
Decision Trees as Rules Computers Follow
Once students have built a tree by hand, introduce the connection: this is how some computers make decisions. The computer does not know what a dog is. It follows rules — and the rules were built by people who looked at lots of examples. If the rules are good, the computer makes good decisions. If the rules are bad, it makes bad ones.
This framing gives students their first accurate intuition about AI: it is a rule-follower, not a thinker. And the rules came from somewhere.
Agents and Rules at the Elementary Level
The agent concept from Topic 6a is accessible at K-5 through a simple framing: an agent is anything that watches and responds. A thermostat watches the temperature and responds by turning the heat on or off. A traffic light watches a timer (or sensors) and responds by changing colors. A voice assistant watches what you say and responds with an answer.
Ask students: what is the agent watching? What can it respond with? What rules does it follow? This three-question framework applies to every agent they will ever encounter and gives them a tool for looking at AI systems analytically rather than magically.
An Unplugged Activity: Be the Robot
One student plays the "robot." Give them a simple rule card: "If someone raises their hand, say 'Yes?'. If someone holds up a red card, say 'Stop.' If someone holds up a blue card, say 'Go.' Otherwise, do nothing." Other students test the robot. Can they predict what it will do? Can they find inputs the rules do not cover?
What the robot cannot do is at least as important as what it can. When a student asks it a question not on the rule card, it is stuck. This is one of the fundamental limits of rule-based AI — and elementary students can discover it on their own through play.
Connections to the Broader K-5 CS Curriculum
- Algorithms: Decision trees are algorithms — step-by-step procedures for reaching a conclusion. Building one reinforces the algorithmic thinking strand that runs through K-5 CS education.
- Data and analysis: Sorting animals by attributes is the same operation as sorting data by attributes. The CS vocabulary and the data literacy vocabulary are the same thing here.
- Digital citizenship: "Computers follow rules that people made" is the foundation for any meaningful conversation about AI fairness, safety, or responsibility. Students who hold that mental model are better positioned for every digital citizenship conversation that follows.