📖 Chapter 4 of 9 · AI For New Learner Advance 1 (Multi Indian Language) · Class 7
Chapter 4: Chatbots & Large Language Models
In this chapter
- What is an LLM? How does ChatGPT really work?
- Zara from Ahmedabad — the art of asking AI well.
- Priyanka from Bhubaneswar — AI as study buddy.
- Indian LLMs: Bhashini, Sarvam, Krutrim.
- Prompt engineering — a real new career.
LLM = Large Language Model
An LLM is an AI trained on BILLIONS of words from books, websites, articles, code. It learns ‘which word usually comes next’. So when you ask a question, it predicts the most likely useful answer — word by word, very fast. ChatGPT, Claude, Gemini, and India’s own Sarvam — all are LLMs.
Did you know?
India launched its own LLM called ‘Sarvam-1’ in 2024 — trained mostly on INDIAN languages. Krutrim, by Ola, was India’s first AI unicorn (billion-dollar) startup. The next big LLM may be made by an Indian — maybe by YOU one day!
Story 1: Zara’s Chatbot Magic
From: Ahmedabad, Gujarat · Zara, age 11
Zara studies in Ahmedabad. Her science project was on water conservation. She opened her father’s chatbot app and typed: “Tell me about water.”
The chatbot wrote 2 pages about water — H2O molecules, ocean tides, water cycle. Useful, but generic. Not what Zara needed for her local project.
Her elder sister Heena saw this. “Bahen, you must TELL the AI who you are and what you really want. AI is like a clever uncle — if you ask vaguely, you get a vague answer.”
Zara tried again: “I am an 11-year-old in Ahmedabad. Write 5 SIMPLE ways my family can save water at home, with examples that work in Gujarat. 4 sentences each.”
WHOA. The chatbot wrote perfectly: close tap while brushing, reuse RO reject water, bucket bath, fix leaky taps, rainwater harvesting in Gujarat (where rain is precious). Zara learned: HOW you ask matters more than WHAT you ask. This skill is called ‘prompt engineering’ — and people are getting paid lakhs for it!
Story 2: Priyanka’s AI Study Buddy
From: Bhubaneswar, Odisha · Priyanka, age 12
Priyanka studies in a school in Bhubaneswar, Odisha. The school got a new AI tutor app last month — free for all students. It can answer doubts, give practice questions, even explain in Odia language!
One day Priyanka was stuck on a maths problem: 248 ÷ 8 = ?. The AI didn’t give the answer. It gave HINTS instead: Hint 1 — try long division. Hint 2 — 8 × 30 = 240. Hint 3 — 248 − 240 = 8. What’s 8 ÷ 8? You can do it!
Priyanka solved it. “31!” she shouted. The next day she helped her classmate the SAME way. “Don’t ask AI for the answer. Use it like a guide — that way YOU learn.”
Now the whole class uses AI as a study buddy. Marks have improved. Teachers say: “AI helps with practice. We help with understanding. Together — we make better students.” That is NEP 2020 in action — joyful, hands-on, and personalised learning.
Solved illustrations
Master the art of LLMs
1. What does ‘LLM’ really mean?
Show answer
Answer: Large Language Model. LARGE = trained on billions of words. LANGUAGE = it works with text. MODEL = a pattern-finding system that predicts which words go together. ChatGPT, Claude, Gemini are all LLMs.
2. Why is asking AI the RIGHT way so important?
Show answer
Answer: AI does what you ask. If you ask vaguely, you get vague answers. The skill of asking well — clear, specific, with context — is called ‘prompt engineering’. It is a real career, with people earning 10-50 lakh rupees yearly. Practice it!
3. Can an LLM ‘invent’ something new?
Show answer
Answer: Tricky. LLMs do not really invent — they REMIX what they learned. Like a chef mixing known ingredients in new ways. True invention (a new theory, a new song style) still needs human imagination. But LLMs help by generating LOTS of ideas to choose from.
4. What are some Indian LLMs?
Show answer
Answer: Bhashini (translation), Sarvam-1 (multilingual), Krutrim by Ola, AI4Bharat by IIT Madras. India is building LLMs trained on INDIAN data — which understand cricket, sambar, Diwali better than American LLMs.
5. Priyanka’s AI gave HINTS not answers. Why is this smart?
Show answer
Answer: Because the goal of school is to make YOU smart — not to make AI solve your problems. Hint-based AI is like a kind teacher: helps you think, doesn’t do your work. This is real future learning.
How a chatbot works
From your typing to the AI’s answer: the LLM reads your prompt, then predicts the most useful next word, and the next, and the next — building the reply one word at a time, very fast.
Try this at home
Open a free chatbot with a parent. First ask: “Tell me about Diwali.” Then ask: “I am 11 in Mumbai. Write a 4-line poem about Diwali in simple English.” See how the better prompt = better answer!
Questions — Chapter 4
Become a prompt engineer
A. Fill in the blanks
- LLM stands for .
- India’s translation AI is called .
- The skill of asking AI well is called engineering.
- Sarvam and Krutrim are Indian .
- A chatbot predicts the next — one at a time.
B. Rate these prompts: WEAK or STRONG?
- “Tell me about mangoes.” →
- “I am 11 in Lucknow. Write 5 fun facts about Alphonso mangoes for my school project.” →
- “Help.” →
- “Write a 4-line Hindi poem about my mother’s love, simple words.” →
C. Short Answer
- Why is Priyanka’s AI tutor smart for giving HINTS, not answers?
- Name 2 Indian LLMs and what they do.
- Write your own STRONG prompt for: planning your birthday party.
Brain Buster
Write 5 strong prompts. Test them with a parent’s permission. Note: which AI gave the best answers?
Check your understanding — Chapter 4
Tap an answer to check it instantly. These are for practice — nothing is saved or sent.
Q1. What does 'LLM' stand for?
- Long Letter Mail
- Large Language Model
- Light Learning Mode
Show answer
B. Large Language Model
Q2. Why is asking AI the RIGHT way so important?
- It isn't important
- Clear, specific prompts get useful answers — this skill is prompt engineering
- AI only works in English
Show answer
B. Clear, specific prompts get useful answers — this skill is prompt engineering
Q3. Which of these are Indian LLMs / AI tools?
- Only ChatGPT
- Bhashini, Sarvam, Krutrim, AI4Bharat
- None exist
Show answer
B. Bhashini, Sarvam, Krutrim, AI4Bharat
Q4. Priyanka's AI tutor gave HINTS, not answers. Why is that smart?
- It was broken
- Because the goal is to make YOU smart, not to make AI do your work
- Hints are faster to type
Show answer
B. Because the goal is to make YOU smart, not to make AI do your work