Term 1 · How AI Really Learns (Machine Learning) · Lesson 1 of 30All lessons
AI CourseSyllabax AI · Class 7Term 1 · How AI Really Learns (Machine Learning)
Term 1 · How AI Really Learns (Machine Learning) · Lesson 1

From Rules to Learning: A Recap and a Leap

6 min read30 practice questions15-question quiz

We see how computers moved from fixed rules to learning from examples, just like we learn new games by playing.

A friendly opening + what we'll learn

Hello, young explorer! Today we will travel from the world of rules to the exciting land of learning. We will see how computers used to need exact instructions for everything, and how they now can learn by looking at many examples – just like you learn to play cricket by watching friends. By the end, you will know why AI can guess your favourite songs on YouTube and how it gets better each day.

Core idea explained simply

Imagine you have a toy robot that can only follow rules you write. If you tell it "if the ball is red, move left; if blue, move right", it will always do that. But what if a new colour appears? The robot gets stuck. In machine learning, instead of giving every rule, we give the robot many pictures of balls and tell it which way to move. The robot looks for patterns and creates its own rule‑book inside its model. Now it can handle new colours it never saw before!

✦ The Big Idea

Computers can move from fixed, hand‑written rules to learning patterns from many examples, making them smarter and more flexible.

Going a little deeper

Think about your school lunch (tiffin). A rule‑based system would need a line for every possible item: "if there is samosa, put it in the left box; if there is fruit, put it in the right box". But what if your mom adds a new snack like a cheese roll? The rule system would need a new rule. A learning system looks at many past tiffins, sees where each item usually goes, and then decides on its own where to place the new snack. It learns the pattern "snacks that are small and round go left" without us writing each rule.

◆ Worked Example 1
  1. List 5 tiffin items: samosa, apple, banana, sandwich, chocolate.
  2. Write the rule: if the item is sweet, put in left box; if salty, put in right box.
  3. Apply the rule: samosa → right, apple → left, banana → left, sandwich → right, chocolate → left.
  4. New item appears: cheese roll (salty). Using the rule, we put it in the right box.
  5. In learning, we would show many past items with their boxes, let the computer find the pattern, and it would also put cheese roll in the right box because it learned "salty = right".
◆ Worked Example 2
  1. You speak to a voice assistant and say, "Play my favorite song".
  2. The assistant has no rule for "favorite"; it looks at your listening history (training data).
  3. It notices you often play a Bollywood song after school.
  4. It predicts that the song you want is that Bollywood track and plays it.
  5. Next week you start listening to a new pop song after school. The assistant adds this to its data, learns the new pattern, and next time it may suggest the pop song instead.
📖 AI in the Real World

Riya loves watching cartoons on YouTube. At first, she saw only the shows she searched for. After a few weeks, YouTube started showing her new cartoons that other kids her age liked. The AI looked at the videos she watched, the ones her friends liked, and the time of day she watched. It learned a pattern and began recommending shows she enjoyed without her asking. One day, the AI even suggested a new cricket highlight video just before the big match, and Riya felt like the app knew her favorite sport!

🧪 Try It Yourself
  1. Gather 10 colored beads (red, blue, green).
  2. Write a simple rule: if bead is red or green, place in bowl A; if blue, place in bowl B.
  3. Sort the beads using the rule.
  4. Now, without looking at the rule, try to guess which bowl a new orange bead should go to by looking at the pattern of colors you already sorted.
  5. Discuss with a friend: Did the pattern help you decide? How is this similar to how AI learns?
💡 Remember
  • Rules are exact instructions.
  • Learning uses many examples.
  • A model stores the learned pattern.
  • Predictions are the computer’s guesses based on what it learned.
📝 Homework
  1. With a parent, list 5 daily activities you do (e.g., brushing teeth, eating breakfast).
  2. Write a simple rule for each (e.g., if it is morning, eat cereal).
  3. Watch a short video about a smart assistant (like Siri) and note one way it learns from you.
  4. Share with your family how the assistant’s learning is different from the rules you wrote.
✦ Rule✦ Training Data✦ Model✦ Prediction✦ Learning✦ Algorithm