550+ Arizona students coded with us in 2025. Help us reach more

Classroom // LESSON_KIT

Lessons you can
teach tomorrow

Free, classroom-tested lessons in AI and computer science. No prior CS background needed, and nothing here asks you to give up a unit you already have planned.

Free

No account, no license, no cost to your school.

Classroom-tested

Both lessons come out of our own Arizona classrooms.

Standards-friendly

Built to fold into the fraction, data and science work already on your plan.

The lessons

2 available
LESSON_01 Grades 6-12

Teachable Machine

Train a model live, in the browser, on sticky notes

Students draw two simple shapes on sticky notes, capture twenty training photos of each, train a model and test it live against their own camera. The lesson lands because the model fails in front of them: they see for themselves how the amount and variety of training data changes what the machine gets right.

What you need

  • Two sticky notes per student
  • Dark washable markers
  • A computer with camera access

Runs on

teachablemachine.withgoogle.com

Open the tool

How it runs

  1. 01 Each student draws two distinct simple shapes, one per sticky note.
  2. 02 Open Teachable Machine and start an Image Project on the Standard Model.
  3. 03 Label two classes and capture 20 photos of each shape.
  4. 04 Train the model, then test it live against the camera feed.
  5. 05 Ask the class what happens when no shape is held up at all.

Ask the class

  • How accurate is the model?
  • What would make it better?
  • How do the amount and the variety of the data change the result?

If you have more time

  • Recapture at 60+ photos per shape from varied angles.
  • Add 20+ background photos as a default class, then push for 90% accuracy across all three.
  • Grade 9+: embed the trained model on an Arduino with an OV7670 camera.
Written by Diana Lee Guzman Free to use and adapt
LESSON_02 Grades 3-8

Smart Snacks

How AI learns from data

A snack preference vote turns into a lesson on prediction. Students collect their own class data, write it as fractions, and use it to predict what the next student will pick. It is designed to sit inside the fraction and probability work a class is already doing rather than to take a period away from it.

What you need

  • Paper or a whiteboard
  • No devices required

Runs on

Paper and a class vote

Open the tool

How it runs

  1. 01 Run a snack preference vote across the whole class.
  2. 02 Record the tally and write each result as a fraction of the class.
  3. 03 Use the fractions to predict what the next student will choose.
  4. 04 Test the prediction against a real student, then discuss why it held or broke.

Ask the class

  • What does the data say is most likely?
  • How confident should we be with a class this size?
  • What would a prediction built on a bigger class get right that ours did not?

If you have more time

  • Re-run the vote with a second class and compare the two sets of fractions.
  • Discuss where a real recommendation system gets its data from.
Written by Diana Lee Guzman Free to use and adapt

Want more than two lessons?

We train educators to teach computer science and AI over a seven-week cohort, and we build scope and sequence with districts against the Arizona CS standards.

Teacher & district resources Ask us a question