🐱 Train an Animal-Spotter
Teach the computer to tell two animals apart by showing it pictures. Then play tricks to find out where it gets muddled.
← All AI Maker projects
🔎 What you'll discover about AI
When you show the computer lots of examples, it learns the patterns that make one animal look different from another. The examples you give it are called its training.
🛠 AI helpers you could use
- Teachable Machine by Google — ask a grown-up to open this with you
- Or use your favourite "teach the computer" tool
Tip: you can always use your favourite AI helper instead — just keep a grown-up nearby.
🪜 How to build it
Day 1
- Open Teachable Machine and start a new Image project.
- Pick two animals, like cats and dogs. Make a group (a "class") for each one.
- Show the camera lots of pictures or toys for each animal so it has plenty of examples.
- Press Train, then test it: hold up an animal and see what it guesses.
Day 2
- Try to trick it with hard examples, like a dog photo from far away.
- If it gets confused, add more example pictures and train it again.
- Write down what made it better and what still fooled it.
🧪 Researcher's question
What happens if you only give it three pictures of one animal but thirty of the other? Does it start guessing one of them more often? Why might that be?