Three icon sets, Data Tally, Held-out markers
Open mission materials →
Concept art made with Gemini 3.1 Flash Image← Mission map
Mission 2 of 12Teachable Machine45 to 55 minutes
Turbo Kart City Rush: Badge Garage
The stunt kart needs balanced BOOST, SHIELD, and JUMP examples before the city-rush clock starts.
Today's AI power
Labels, examples, and balance
AI learns from examples. Balanced, varied examples help it recognize new cards and props.
Start here
Install three parts with held-out scans and finish a short stunt combo.
All three classes begin with equal example counts.
Image Model v1 and the full kart kit
See how this game works
Game loop preview
See the power before you build it.
- 1SignalBalanced badge examples
- 2AI moveThe model predicts a badge
- 3Your ruleThe exact game logic checks what happens next.
- 4Game stateSCANNING
Your mission
One step at a time. One AI power to unlock.
Step 1 of 9Story
Story
Meet the mission problem.
Story signal: The stunt kart needs balanced BOOST, SHIELD, and JUMP examples before the city-rush clock starts.
Do this now
- 1LOOK
Spot the problem in the opening story and tell your crew what needs to change.
Goal
Know what winning looks like.
Your choice: Choose which weak class receives more useful variety.
Watch this change: Balanced, varied data makes the weak class more recognizable in held-out tests.
Do this now
- 2CHOOSE
Win by proving this: All three classes begin with equal example counts.
Get ready
Gather the props and get your grown-up guide.
Do this now
- 3CHECK
Open the Mission 2 materials on screen or print them, then gather the listed props.
- 4CHECK
Use cards, symbols, or toys on a plain tabletop and keep the camera focused on that prop zone.
Predict
Guess first, then test.
Do this now
- 5CHOOSE
Name the three classes and predict which two may look similar.
Build
Make the AI power in the Google tool.
Do this now
- 6MAKE
Follow the Google tool card below. Complete each step in order, then bring the visible result back here.
Google tool step
Open the tool, make the power, then bring one result back.
OPEN TOGETHER
Teachable Machine
Adult and child operate together
Build thisImage project
Save asImage Model v1
Do these steps
- Open an Image Project.
- Create the three exact labels from the Label Map.
- Collect balanced examples in varied conditions.
- Train Image Model v1.
- Test each held-out example before adding it.
You are done whenThree prediction bars respond to the held-out icons and expose the weakest class.
Need setup help?
- Bring
- One three-icon set
- Completed Data Tally
- Three held-out examples
- Exact label map
- Next move
- Record all held-out results and choose one evidence-based data improvement.
- If stuck
- Compare labels, counts, and backgrounds before collecting any more examples.
Bring it back
Return with the result you can see.
Do this now
- 7CHECK
Save the label map and model version name. Save the exact mission artifact as Image Model v1, then return to this card.
Glitch test
Make a funny mistake on purpose.
Funny glitch: The kart grows three shields and one tiny wheel.
Do this now
- 8TEST
Remove half the examples from one practice stack. Watch how an unbalanced set can create funny mixed-up powers.
Boss round
Change one thing and retry.
Game move: Kart upgrades, launch ramps, and part-collection streaks
Harder remix: Write a short data card describing class balance, variation, possible shortcuts, and known limits.
Do this now
- 9REMIX
Choose one evidence-based change, run the same test again, and compare what changed.
- 10TEST
Run the boss round: Image Model v1 and its label map are saved.
Explain
Say what the AI did and prove it.
Power reward: Data Designer · Image Model v1 and the full kart kit
AI learns from examples. Balanced, varied examples help it recognize new cards and props.
AI does not magically know the classes. Better data choices reduce shortcuts and make behavior easier to test.
Do this now
- 11SAY
Who chose the labels, and how did balanced, varied examples help the model?
- 12SAVE
Complete the mission proof and save only the named artifact: Image Model v1.
What counts as finished?
Who chose the labels, and how did balanced, varied examples help the model?
- All three classes begin with equal example counts.
- At least two conditions vary within each class.
- A held-out example is tested for each label.
- Image Model v1 and its label map are saved.
Need help or want a harder challenge?
Make this step easier
Use a 10-box tally strip for each class and copy the same four camera poses with every card.
Say it this way: The AI noticed ___. The exact game rule ___. I know because ___.
Try the deeper remix
Write a short data card describing class balance, variation, possible shortcuts, and known limits.
- Record the inputs and outputs.
- Compare two versions or conditions.
- Name the constraint or exact rule.
- Design one fair remix test.
Everyone proves the same AI idea: All three classes begin with equal example counts. · At least two conditions vary within each class. · A held-out example is tested for each label. · Image Model v1 and its label map are saved.
Quick power check