📖 Chapter 2 of 9 · AI For New Learner Advance 1 (Multi Indian Language) · Class 7
Chapter 2: What AI CANNOT Do
In this chapter
- Things AI is GREAT at — and where it FAILS.
- Neel from Kolkata shows the limits of AI.
- Maya from Shillong scans real nature with AI.
- Why human teachers, doctors, parents are still needed.
- When NOT to trust an AI’s answer.
The Big Idea
AI can BEAT humans at many tasks — chess, X-ray reading, fast translation. But there are still things AI CANNOT do: feel emotions, understand WHY things matter, make moral choices, or know what is right when nobody told it. This is why AI works BEST when paired with humans.
Try this at home
Ask any chatbot: “How does it feel when your favourite team loses?” Compare its answer with what YOU would say to a sad friend. What is missing?
Story 1: Neel’s Bridge of Truth
From: Kolkata, West Bengal · Neel, age 12
Neel lives near the old Howrah Bridge in Kolkata. His grandfather, Dadu, was an engineer who built bridges. One Sunday Dadu was very sad — his old friend had passed away. He sat quietly by the river.
Neel thought, “Let me use AI to cheer up Dadu!” He opened a chatbot: “My grandfather is sad. Write something to make him happy.” The chatbot wrote a nice paragraph: ‘Loss is part of life. Time heals everything. Stay positive.’
Neel read it to Dadu. Dadu smiled gently. “Beta, the words are correct. But they are empty. The chatbot does not know my friend. It does not know me. It does not know what we did together for 60 years.” Dadu hugged Neel. “Your hug just now — THAT did more than 100 chatbot lines.”
Neel learned a big lesson that day. AI can WRITE about love. It cannot LOVE. AI can WRITE about loss. It does not FEEL loss. There are some things only humans can do — and they matter most when life gets hard.
Story 2: Maya’s Nature Quest
From: Shillong, Meghalaya · Maya, age 11
Maya lives in the cool hills of Shillong. Her school took her class on a trek to see the famous Living Root Bridges. Maya carried her father’s old mobile phone with a Nature ID app — an AI that identifies plants and birds from photos.
The app was amazing! Maya pointed it at flowers — “Hibiscus, 99% sure.” At a bird — “Indian Robin.” At a strange root — “Banyan tree root.” In one hour Maya identified 23 species!
But then a strange leaf. The app said “Unknown — try a clearer photo.” Maya took five photos. Still “Unknown.” Their tribal guide, Kyrmen-da, looked and smiled — “That is the local Pajan leaf. We Khasi people use it for medicine for 500 years. The AI does not know our local plants.”
Maya realised — AI is trained on what data it sees. The Western internet has lots of English-named plants. Local Indian and tribal knowledge is missing from many AIs! Her project at school: “Add 30 Meghalaya plants to AI training data — make AI more Indian!”
Solved illustrations
Understanding the limits of AI
1. Can AI ‘feel’ emotions like sadness or joy?
Show answer
Answer: No. AI can RECOGNISE emotions (face expression analysis) and TALK about emotions (using words from training data). But it does not actually FEEL anything. Feelings come from being alive — having a body, family, hopes, fears. AI has none of these.
2. Why did the chatbot’s reply not help Dadu?
Show answer
Answer: Because the chatbot did not know Dadu’s friend, did not know their 60 years together, and did not understand what was lost. Comfort comes from a person who SHARES the bond. AI can give general words, not personal love.
3. Why did the Nature ID app fail for the Pajan leaf?
Show answer
Answer: Because AI only knows what it was trained on. If 99% of plant photos online are common Western plants, the AI cannot identify rare local Indian plants. This is called a ‘data gap’.
4. What is the lesson of Maya’s story for India?
Show answer
Answer: India must put MORE Indian data into AI — local languages, tribal knowledge, regional crops, Indian faces. Otherwise our AI will know more about American plants than about Pajan leaves. We must build AI for OUR country.
5. When should you NOT trust an AI’s answer?
Show answer
Answer: When the topic is — (a) emotional/personal (b) local/cultural (c) very new (after the AI was trained) (d) life-critical (medical, legal). In these, check with a real human expert first.
Did you know?
When ChatGPT was first launched in 2022, it could not name the current Prime Minister of India correctly — because its training data was older. AI is trapped in the past. Always check important facts on real news websites.
Questions — Chapter 2
Think carefully — these have no single right answer
A. True or False
- AI can feel sad when its team loses. ( )
- The Nature ID app knows every plant in the world. ( )
- AI is trained on data — if data is missing, AI fails. ( )
- Comforting a sad person needs more than nice words. ( )
- India does not need its own AI training data. ( )
B. Match the situation with what is needed
| Situation | What is needed |
|---|---|
| Lost a pet | AI translator |
| Found a strange plant | AI image-search |
| Got a math problem | AI calculator |
| Need to translate Hindi | A real friend |
C. Discuss & Write (3-4 sentences)
- Why does Maya want to add Meghalaya plants to AI data? Why does it matter for India?
- Give one example of a problem where you would prefer a real human, not AI.
- AI cannot feel — but it can WRITE about feelings. Is this dangerous or useful? Why?
Brain Buster
Make a list of 5 things that ONLY humans can do — and discuss with your class.
Check your understanding — Chapter 2
Tap an answer to check it instantly. These are for practice — nothing is saved or sent.
Q1. Can AI 'feel' emotions like sadness or joy?
- Yes, fully
- No — it can recognise and talk about emotions, but not feel them
Show answer
B. No — it can recognise and talk about emotions, but not feel them
Q2. Why did the Nature-ID app fail on the Pajan leaf?
- The camera was broken
- It was only trained on common plants, not rare local ones
- The leaf was too small
Show answer
B. It was only trained on common plants, not rare local ones
Q3. When should you NOT fully trust an AI's answer?
- Never, AI is always right
- For emotional, local, very new, or life-critical topics
- Only on weekends
Show answer
B. For emotional, local, very new, or life-critical topics
Q4. What is the lesson of Maya's story for India?
- India should stop using AI
- India must add more Indian data — languages, crops, faces — to AI
- AI is useless in villages
Show answer
B. India must add more Indian data — languages, crops, faces — to AI