Class 9 · Unit 1: The AI project cycle, and ethics · Lesson 1.1 · 40 min

Three domains of AI, and the SDGs

Three projects. Bilal says they're all the same amount of work. One question proves him wrong.

Today you will: Place any AI by what it looks at · Name an SDG AI could serve here · Name the six stages of the AI project cycle

Story

The science exhibition is in October, and the AI club can enter one project. Three ideas are on the board:

A checker that looks at a photograph of an apple leaf and says whether it has scab.
A predictor that takes last year's rainfall and yield figures and estimates this season's crop.
A sorter that reads the slips from the suggestion box and groups them by what they are about.

Bilal: “They're all AI, so they're all the same amount of work. Let's do all three.”

Hiba is already writing on the board. Three columns, one word at the top of each: pictures · numbers · words.

Hiba: “They're not the same work at all. Look at what each one has to look at.”

Watch

The three domains, and the cycle in CBSE's words — — scan code 9.1.1 in the printed book.

Warm up your fingers

Skill: the symbols code is made of (6 min)

On typing.com: the symbols lessons — ( ) : " = _ [ ].
Type five times, eyes on the screen. It's real Python; you'll run it in Unit 5:

print("Salaam, Srinagar")
Every bracket and quote must close. Goal this term: 32–38 WPM at 94%.

On the laptop

You need: LibreOffice Calc · domains.ods · sdg-list.odt · a browser (no accounts)

Mission 1: Sort fifteen applications (11 min, in pairs)

☐1
Open domains.ods. For each application, put an x under every domain that applies: Data · Computer Vision · Natural Language.
☐2
In column E, write in four words what the machine looks at — rows of numbers, a photo, a sentence, a voice.
☐3
In column F, name one SDG it could serve (see sdg-list.odt). If it serves none you can defend, say so.
☐4
Predict: three applications need two domains. Guess which, then check and highlight them.

Mission 2: Meet the three domains (8 min, in threes)

☐5
Take turns on three games — no account needed:
- Rock, Paper, Scissors (next.rockpaperscissors.ai) learns your pattern — data.
- Quick, Draw! (quickdraw.withgoogle.com) recognises drawings — computer vision. Draw objects only, never people.
- Semantris (research.google.com/semantris) matches words by meaning — natural language.
☐6
After each, write one line in sheet Games: what did it look at, and what did it get wrong?

Finished early?

Invent a sixteenth application that needs all three domains.

No laptop today?

Card sort: fifteen slips, three headings on the board; pairs place slips and defend two. Then act the domains out — read a sentence aloud, hold up a picture — and compare what each asks you to look at.

Now you know

The three AI domains, sorted by what the machine takes in: numbers, pictures, or words.

One question places any AI: what does the machine look at — numbers, pictures or words?
Overlap is normal. A voice assistant is NLP and data. Name the main domain first.

AI can serve Sustainable Development Goals such as Zero Hunger, Good Health, Quality Education, Clean Water and Climate Action — and strain another, such as Reduced Inequalities, when a service needs a smartphone.

The SDGs are 17 goals the world's governments agreed in 2015, to reach by 2030.
Ask two questions: which goal does this project serve — and which does it strain?

The six CBSE 417 stages — problem scoping, data acquisition, data exploration, modelling, evaluation, deployment — mapped to the names learned in Class 8.

Every project runs the AI project cycle: problem scoping → data acquisition → data exploration → modelling → evaluation → deployment. It's Class 8's journey with CBSE's labels.

Debate it

The AI club can enter one project in October.

Which of the three ideas needs the most data the club doesn't have yet?
The suggestion slips are in English, Urdu and Kashmiri. Why is that harder than the leaf checker?
Which would you choose for six weeks — and what would you say to the students whose idea lost?

Check yourself — practice, not a test

1. Match each application to its main domain:
A checker that reads a photo of a leaf · Predicting this season's yield from rainfall figures · Grouping complaint slips by topic · A number-plate reader at a gate
Match with: Data · Computer Vision · Natural Language Processing
2. Which question places an AI system in its domain fastest?
- ○ How expensive was it?
- ○ What does the machine look at — numbers, pictures or words?
- ○ Who built it?
- ○ Does it run on a phone?
3. Put the six stages of the AI project cycle in CBSE 417's order:
- 1. Data acquisition
- 2. Evaluation
- 3. Data exploration
- 4. Deployment
- 5. Problem scoping
- 6. Modelling
4. An AI system belongs to exactly one domain and never more than one.
- ○ True
- ○ False
5. Working with written or spoken human language is the domain called natural language ______.
6. A checker that spots apple scab early, so a grower does not lose the harvest, most directly serves which Sustainable Development Goal?
- ○ Goal 13, Climate Action
- ○ Goal 2, Zero Hunger
- ○ Goal 4, Quality Education
- ○ Goal 6, Clean Water and Sanitation

Remember

Ask what the machine looks at: numbers, pictures or words.

Story

The science exhibition is in October, and the AI club can enter one project. Three ideas are on the board:

A checker that looks at a photograph of an apple leaf and says whether it has scab.
A predictor that takes last year's rainfall and yield figures and estimates this season's crop.
A sorter that reads the slips from the suggestion box and groups them by what they are about.

Bilal: “They're all AI, so they're all the same amount of work. Let's do all three.”

Hiba is already writing on the board. Three columns, one word at the top of each: pictures · numbers · words.

Hiba: “They're not the same work at all. Look at what each one has to look at.”

On the laptop

You need: LibreOffice Calc · domains.ods · sdg-list.odt · a browser (no accounts)

Mission 1: Sort fifteen applications (11 min, in pairs)

☐1
Open domains.ods. For each application, put an x under every domain that applies: Data · Computer Vision · Natural Language.
☐2
In column E, write in four words what the machine looks at — rows of numbers, a photo, a sentence, a voice.
☐3
In column F, name one SDG it could serve (see sdg-list.odt). If it serves none you can defend, say so.
☐4
Predict: three applications need two domains. Guess which, then check and highlight them.

Mission 2: Meet the three domains (8 min, in threes)

☐5
Take turns on three games — no account needed:
- Rock, Paper, Scissors (next.rockpaperscissors.ai) learns your pattern — data.
- Quick, Draw! (quickdraw.withgoogle.com) recognises drawings — computer vision. Draw objects only, never people.
- Semantris (research.google.com/semantris) matches words by meaning — natural language.
☐6
After each, write one line in sheet Games: what did it look at, and what did it get wrong?

Finished early?

Invent a sixteenth application that needs all three domains.

No laptop today?

Now you know

One question places any AI: what does the machine look at — numbers, pictures or words?

Overlap is normal. A voice assistant is NLP and data. Name the main domain first.

The SDGs are 17 goals the world's governments agreed in 2015, to reach by 2030.

Ask two questions: which goal does this project serve — and which does it strain?

Every project runs the AI project cycle: problem scoping → data acquisition → data exploration → modelling → evaluation → deployment. It's Class 8's journey with CBSE's labels.

Debate it

The AI club can enter one project in October.

Which of the three ideas needs the most data the club doesn't have yet?
The suggestion slips are in English, Urdu and Kashmiri. Why is that harder than the leaf checker?
Which would you choose for six weeks — and what would you say to the students whose idea lost?

Check yourself — practice, not a test

1. Match each application to its main domain:

A checker that reads a photo of a leaf · Predicting this season's yield from rainfall figures · Grouping complaint slips by topic · A number-plate reader at a gate
Match with: Data · Computer Vision · Natural Language Processing

2. Which question places an AI system in its domain fastest?

○ How expensive was it?
○ What does the machine look at — numbers, pictures or words?
○ Who built it?
○ Does it run on a phone?

3. Put the six stages of the AI project cycle in CBSE 417's order:

1. Data acquisition
2. Evaluation
3. Data exploration
4. Deployment
5. Problem scoping
6. Modelling

4. An AI system belongs to exactly one domain and never more than one.

○ True
○ False

5. Working with written or spoken human language is the domain called natural language ______.

6. A checker that spots apple scab early, so a grower does not lose the harvest, most directly serves which Sustainable Development Goal?

○ Goal 13, Climate Action
○ Goal 2, Zero Hunger
○ Goal 4, Quality Education
○ Goal 6, Clean Water and Sanitation