Your First Simple AI Experiment
Get hands-on experience with a simple, no-code AI tool to understand the basic input/output process.
Time to Get Hands-On!
This step involves actually building and training your first AI model. Don't worry - it's completely beginner-friendly and requires no coding!
Meet Google Teachable Machine
What is Teachable Machine?
A free, web-based tool by Google that lets you train machine learning models without any coding. You can teach it to recognize images, sounds, or poses!
Image Recognition
Teach AI to recognize objects, animals, or gestures
Audio Classification
Train AI to recognize different sounds or voices
Pose Detection
Recognize body positions and movements
Let's Build Your First AI Model!
We'll create an image classifier that can tell the difference between cats and dogs. Follow these steps:
Open Teachable Machine
Go to teachablemachine.withgoogle.com and click "Get Started"
💡 Tip: No account needed! It works directly in your browser.
Choose "Image Project"
Select "Image Project" and then "Standard image model" to start building your classifier.
Create Your Classes
Rename "Class 1" to "Cat" and "Class 2" to "Dog". You can add more classes if you want!
⚠️ Important: Each class needs at least 10 examples to work well.
Add Training Images
For each class, click "Upload" and add 10-20 images. You can:
- Upload images from your computer
- Use your webcam to take photos
- Search for images online (make sure they're free to use)
Train Your Model
Click "Train Model" and wait for the magic to happen! This usually takes 1-2 minutes.
✨ What's happening: The AI is learning patterns from your images!
Test Your AI
Once training is complete, test your model by uploading new images or using your webcam. Watch as your AI predicts whether it's seeing a cat or dog!
Understanding What Just Happened
🧠 Pattern Recognition
Your AI learned to recognize patterns that distinguish cats from dogs - things like ear shape, facial features, body proportions, and fur patterns.
📊 Confidence Scores
Notice the percentage scores? That's how confident your AI is in its prediction. Higher percentages mean the AI is more certain about its answer.
🎯 Training vs. Testing
The images you uploaded were for training. When you test with new images, you're seeing how well your AI can generalize what it learned to new, unseen data.
Try These Fun Experiments!
🍎 Food Classifier
Train AI to recognize different fruits, vegetables, or types of food.
✋ Hand Gesture Recognition
Teach AI to recognize different hand gestures like thumbs up, peace sign, or rock-paper-scissors.
🎨 Art Style Detector
Train AI to distinguish between different art styles or artists' works.
🌤️ Weather Classifier
Create a model that can identify sunny, cloudy, rainy, or snowy weather from photos.