Free AI Course for Kids: Train Your Own AI Model in 30 Minutes

Kids use AI every day: voice assistants, video recommendations, photo filters. This free course helps them understand how it actually works by training their own AI model. It takes about 30 minutes, suits ages 9 and up, and all you need is a laptop with a webcam.

Part 1: How does an AI learn?

Normal programs follow rules a person writes: "if the score is over 90, say Amazing." Machine learning is different. Instead of writing rules, we show the computer lots of examples, and it finds the patterns itself.

Think about how you learned to recognise a cat. Nobody gave you a rule like "four legs plus whiskers". You saw many cats, and your brain learned the pattern. Machine learning works in a similar way:

1. Collect examples (data), like many photos of cats and dogs, labelled.
2. Train a model: the computer adjusts itself until it can tell them apart.
3. Test it on examples it has never seen.

Part 2: Train your own model (free, in the browser)

We'll use Teachable Machine, a free tool from Google that runs in your web browser.

1. Search for Teachable Machine, open it, click Get Started, then Image Project → Standard image model.
2. Rename Class 1 to Thumbs up and Class 2 to Thumbs down.
3. For each class, click Webcam and hold the button to record about 50 pictures of your hand doing that gesture. Move your hand around a little.
4. Click Train Model and wait a few seconds.
5. In the Preview panel, show a thumbs up or a thumbs down. The model tells you what it thinks, and how confident it is.

You just trained an AI.

Part 3: Experiments (this is where the learning happens)

1. Add a third class called Nothing, with pictures of just you and no gesture. Does the model get better or worse?
2. Fool your model: try a different background, dim lighting, or the other hand. When does it get confused?
3. Less data: make a new project with only 5 pictures per class. How accurate is it now?

What you'll discover is the most important lesson in AI: a model is only as good as its data. If all your thumbs-up photos were in a bright room, it may fail in a dark one. Real AI systems have the same problem, which is why they sometimes make unfair or silly mistakes.

Part 4: Using AI responsibly

Talk about these with a parent or friend:

• AI can be wrong, and it can sound confident even when it is. Check important facts.
• Bias: if a model learned from data that left some people out, it may work worse for them.
• Privacy: don't put personal information or photos of other people into AI tools without permission.
• Honesty: if you use AI for schoolwork, follow your school's rules and say when you used it.

Challenges

• Build a Rock, Paper, Scissors model with three classes.
• Try an Audio Project in Teachable Machine that recognises claps and whistles.
• Ready for more? Learn Python with our free Python course. Most real AI is written in Python.

Go further

CodingChamp's live AI & Machine Learning course teaches kids to build image and text classifiers and beginner chatbots, 1:1 with a trainer. Book a free trial class.