A2 Teach a Machine
How do we teach a machine by showing it examples?
Lesson plans ready, classifier tool in development
- Stage
- Builders, ages 7 to 9
- Lessons
- 6 × 40 minutes (4 hours)
- Ends with
- A trained classifier and a fairness note
- Before you start
- A1, or comfort using a mouse or touch screen
Lesson plans ready, classifier tool in development. The lesson plans are complete. The classifier tool for lessons 3 to 6 is still being built. Until it is ready, run those lessons with drawn example cards, and use the Try page for the teaching moment.
For children to do on their own. Every lesson has a step-by-step do-it-yourself version with a self-check. Open the kids’ version.
By the end, children can say
- I can explain AI in my own words.
- I can say why the examples we give a machine matter.
- I can train a classifier with at least 2 classes.
- I can count right and wrong results and improve a model.
- I can spot when a model was trained on examples that leave things out.
- I can present my model, say how well it works and say where it might fail.
About this course
Format. Card games and a classifier tool that trains small models in the browser.
Tools. The classifier tool is planned to run in the browser, so pictures and sounds stay on the device. Pictures should be drawings or objects, never faces.
Key words. AI, data, example, label, train, test, accuracy (right and wrong), fair.
Curriculum tags.
- Computer science: data and algorithms
- Digital literacy
The current national curriculum does not mention AI by name, so each course is tagged to the strands it supports.
Lesson plans
Open a lesson to see the steps. Printing this page opens them all.
Lesson 1: What Is AI?I can explain AI in my own words.
- Recap and hook (8 min). A1 recap, then a 60-second animation on AI examples.
- Teach (8 min). AI is a computer that learns patterns from examples instead of being told every rule.
- Activity (19 min). Match everyday tools to “follows fixed rules” or “learns from examples”.
- Reflect (5 min). One sentence: “AI is…”
Lesson 2: Examples Teach MachinesI can say why the examples we give a machine matter.
- Hook (6 min). The human classifier game: one child guesses “apple or banana” after seeing only 1, then 5, then 10 examples.
- Teach (8 min). Introduce data, label and train.
- Activity (20 min). In groups, build a card set to train a “classmate machine” to sort “healthy snack” and “treat”.
- Reflect (6 min). What happened with few examples? With confusing examples?
Lesson 3: Train a Picture SorterI can train a classifier with at least 2 classes.
- Recap (5 min). Data, label, train.
- Demo (7 min). Draw two shapes (circles and squares) and add them as examples in the tool.
- Build (22 min). Train a classifier on two classes of children’s own drawings or classroom objects; test it with new items.
- Reflect (6 min). What did the classifier get right? What confused it?
Lesson 4: Test ItI can count right and wrong results and improve a model.
- Hook (5 min). Show a classifier that mixes up two items.
- Teach (8 min). Testing: try new examples and count how many are right.
- Activity (22 min). Test with 10 new items, fill a tally chart, add better examples, test again.
- Reflect (5 min). Did the score go up? Why?
Lesson 5: When Examples Are UnfairI can spot when a model was trained on examples that leave things out.
- Hook (5 min). A classifier trained on only red apples fails with green apples.
- Teach (8 min). If a set leaves things out, the machine can be unfair. Ask: “Who or what is missing?”
- Activity (22 min). Fix a lopsided example set by adding the missing kinds, then retest.
- Reflect (5 min). One sentence for a fairness note: “Our classifier was unfair to… until we added…”
The “Only teach the Hill” button on the Try page shows this idea on screen.
Lesson 6: Show and ShareI can present my model, say how well it works and say where it might fail.
- Plan (7 min). A planning sheet: what it sorts, what examples, how I tested it.
- Build (18 min). Finish and test a classifier; write the fairness note.
- Share (10 min). Demo to a partner who tries to trick it.
- Reflect (5 min). Badge and a take-home question: “What is something a machine got wrong for you?”
For grown-ups
- Ask: “How would you teach a machine to tell an apple from a banana?”
- Ask: “What is something a machine got wrong for you?”
- Look at a toy or app together and talk about whether it follows rules or learns from examples.
In every lesson: the “check it” habit
Who made this? What was it trained on? Where else can I check?