📖 Chapter 3 of 9 · AI For New Learner Basic 2 (Multi Indian Language) · Class 5
Chapter 3: How Machines Learn New Things
In this chapter
- How does a computer learn? It is called Machine Learning!
- Training data — showing AI many many examples.
- Patterns — the secret recipe of all learning.
- Meeting Tenzin (Sikkim) and Meera (Rajasthan).
What is machine learning?
New word: Machine Learning
Machine Learning (ML) is the way computers LEARN from examples — without being told every rule. Just like you learn to recognise a mango by seeing many mangoes, a computer learns a mango by seeing thousands of mango pictures.
ML is the brain of AI. Almost every smart machine you see uses Machine Learning to get better and better.
Story 1: Tenzin Teaches the Computer
From: Gangtok, Sikkim · Character: Tenzin, age 10
Tenzin lives in Gangtok with a beautiful view of the Kanchenjunga mountains. In his village there are two animals — yak and dzo (a yak-cow mix). They look very similar! Tourists often confuse them.
Tenzin’s school had a project — make an AI app to tell yak from dzo. With his teacher’s help, Tenzin took 200 photos of yaks and 200 photos of dzos. They fed all 400 pictures (with labels) to a Machine Learning program.
After learning from so many examples, the AI could now look at any new photo and say “yak” or “dzo” correctly. Tenzin was thrilled. “I taught the computer just like Ama (mother) teaches me!” he laughed. The secret of AI learning is — many many examples, just like a child.
Story 2: Meera and the Camel Helper
From: Jaisalmer, Rajasthan · Character: Meera, age 9
In the golden desert of Jaisalmer, Meera’s family runs camel rides for tourists. Some camels are friendly. Some are tired. Some need rest. How do you know which is which?
Meera’s elder brother made a clever app. Every day for one month, he wrote down each camel’s name, walking speed, food eaten, and mood (happy / tired / hungry). After many weeks of data, the AI on the phone learned the PATTERN — “If a camel walks slowly and eats little, it is tired.”
Now the app tells the family which camel can carry tourists and which one needs rest. The camels are healthier. The tourists are safer. Meera says proudly — “Our desert is old, but our ideas are new!”
Solved illustrations
Step-by-step examples of how machines learn
1. What is ‘training data’?
Show answer
Answer: Training data is the bunch of examples we show the AI so it can learn. Tenzin’s 400 photos of yak and dzo are training data. The more examples you give, the better the AI learns.
2. What is a ‘pattern’ in machine learning?
Show answer
Answer: A pattern is a repeating idea the AI finds in the data. For example: ‘slow walking + low food = tired camel’. Once AI learns the pattern, it can use it to take decisions in new situations.
3. Why did Tenzin need both yak photos AND dzo photos?
Show answer
Answer: To learn the difference, AI must see both sides. If it only sees yaks, it cannot tell what is NOT a yak. Showing both classes is called labeled training.
4. Can AI keep learning forever?
Show answer
Answer: Yes, AI can keep improving when we give it new examples. Meera’s brother adds new camel data daily. Over time, the AI’s predictions become smarter and smarter.
5. Can AI learn the wrong thing?
Show answer
Answer: Yes! If we give wrong examples, AI will learn the wrong pattern. Example: if all yak photos are at night and all dzo photos in day, AI might think ‘dark = yak’! So we must give clean, fair data.
How is AI’s learning like your learning?
| ★ You (a child) | ★ AI (a computer) |
|---|---|
| Mother shows you a mango many times | We feed AI thousands of mango photos |
| You taste sweet, smell sweet, see yellow | AI sees colour, shape and size data |
| Your brain forms a memory of ‘mango’ | AI forms a ‘model’ inside its memory |
| Next time, you say ‘mango!’ easily | AI also says ‘mango!’ to new photos |
| If you taste a sour mango, you learn that too | If AI sees a green mango, it learns too |
Fun Brain Game
Imagine you have to TEACH a computer the difference between a cat and a dog. What 3 things would you show in many photos?
Hint: think about ears, tail, sound, size, and the way they walk.
Discuss with your friend. There is no single right answer — that is how AI scientists think every day!
Questions — Chapter 3
Think before you write — there are no wrong tries!
A. Choose the correct option
- Machine Learning means a machine — (a) becomes alive (b) learns from data (c) goes to school
- Tenzin took photos of yaks for training. (a) 2 (b) 200 (c) 2,00,000
- A repeating idea that AI finds in data is called a — (a) festival (b) song (c) pattern
- If we give AI wrong examples, the AI will — (a) refuse (b) learn wrongly (c) become angry
B. Short Answers
- What did Meera’s brother record about the camels for one month?
- Why does AI need many examples, not just one or two?
- Name two animals from this chapter that an AI was trained to recognise.
- How is your learning at home similar to a computer’s learning?
Brain Buster
If you had to train an AI to recognise your school bag, what 5 things would the photos need to show?
Check your understanding — Chapter 3
Tap an answer to check it instantly. These are for practice — nothing is saved or sent.
Q1. Machine Learning means a machine —
- becomes alive
- learns from data
- goes to school
Show answer
B. learns from data
Q2. How many photos of yaks did Tenzin use for training?
- 2
- 200
- 2,00,000
Show answer
B. 200
Q3. A repeating idea that AI finds in data is called a —
- festival
- song
- pattern
Show answer
C. pattern
Q4. If we give AI wrong examples, the AI will —
- refuse
- learn wrongly
- become angry
Show answer
B. learn wrongly