English teaching pack · हिन्दी में गतिविधियाँ

Print or save as PDF using your browser’s Print menu (Ctrl+P / ⌘P). Swipe wide tables sideways to see every column.

CBSE Computational Thinking & AI · 2026–27

Class 7 · Artificial Intelligence
Chapter 1 Teaching Pack

AI Domains and Applications — a print-ready plan for teaching how data, vision and language systems work.

Complete pack Lesson plan Worksheet + answer key Unplugged activity Project rubric Evidence record

Where this chapter sits

Class 7 Computational Thinking & AI has 100 instructional hours across the year. The Artificial Intelligence strand has four chapters; this opening chapter supplies the vocabulary students will need when they later examine industries, visualise data and discuss bias.

#	AI chapter	Learning focus	Planning guide
1	AI Domains and Applications	Data Science, Computer Vision, Natural Language Processing, prediction, datasets and AI versus automation	8 periods
2	AI in Industries	How AI tasks are combined in real services and sectors	4 periods
3	Data Visualization and Analysis	Reading, comparing and communicating patterns in data	4 periods
4	Ethics and AI Bias Awareness	Fairness, bias, privacy and responsible choices	4 periods

Assessment note The CBSE framework asks for continuous, project- and activity-based assessment from Class 6, not a single end test. Use the worksheet to check individual understanding, the activity to observe reasoning, and the rubric-scored project as retained evidence. The period split above is a teaching suggestion, not an official weighting.

Lesson plan · 8 periods

Period 1 — Rebuild the foundation

Put two cases on the board: a school bell that rings at fixed times, and a mail filter that sorts new messages using patterns from earlier examples. Ask what evidence would prove that each one uses AI. Students often call any fast electronic system “AI”; keep returning to the mechanism.

Working distinction Automation carries out a stated rule. An AI model applies patterns learned from data to a new input. A device can contain both. Judge the particular task, not the whole product.

Period 2 — Three domains, three kinds of input

Use the body analogy only as a memory hook: Data Science acts like a brain comparing organised records; Computer Vision acts like eyes handling pictures and video; Natural Language Processing acts like a mouth and ears working with written or spoken language. The analogy is not a definition.

Domain	Typical input	Typical task	Example output
Data Science	Rows of attendance, sales, weather or sensor values	Compare patterns and estimate an outcome	Expected demand next week
Computer Vision	Pixels in an image or frames in a video	Find an object, face, printed word or location	Boxes around vehicles
Natural Language Processing	Typed text or recorded speech	Interpret, translate, summarise or generate language	A translated sentence

Ask students to state both the input and the job before naming a domain. This prevents guessing from brand names.

Period 3 — Predictions have different shapes

Write three outputs: “urgent / routine”, “37 minutes”, and three unnamed groups of similar books. Give the terms only after students see the difference.

Classification selects from categories fixed in advance.
Regression estimates a number on a scale, such as time, price or rainfall.
Clustering discovers groups of similar records when no group labels were supplied.

A prediction is an estimate supported by past patterns, not a promise about the future. Changed conditions or weak data can make it wrong.

Period 4 — What a dataset contains

Draw a small attendance table. One row is a record; each column, such as date or status, is a feature. Contrast that tidy table with a folder of photographs. Both are datasets, but they are organised differently.

Then split ten imaginary records as 6 + 2 + 2. The first portion teaches the model, the second helps people adjust choices during development, and the untouched last portion checks final performance. Use the terms training, validation and test only after the roles are clear.

Period 5 — Computer Vision is more than taking a picture

A camera captures pixels; it does not by itself understand them. Walk through a short pipeline: collect an image, improve unusable quality where possible, notice useful visual features, then locate or categorise what was found. Use a street sketch and ask for the difference between naming a bus and drawing a box around its position.

Text read from a photograph also begins as a vision task. Understanding the extracted words is a separate language task.

Period 6 — NLP works with meaning and context

Write “The match was light”. Ask whether light means “not heavy”, “not dark” or something that burns. Students will ask for context; that need is the lesson. NLP systems use patterns in language data to choose likely meanings, purposes or sentiment, but can still misread sarcasm, mixed languages and unfamiliar names.

Period 7 — Real systems cross domain boundaries

Trace a photographed restaurant review. Vision can turn the photographed lettering into machine-readable text; NLP can judge what the sentence expresses; Data Science can combine many results into a trend. Ask where personal data enters and what should not be collected. Students must learn that a feature can involve more than one domain.

Period 8 — Sort, defend, revise

Run the unplugged activity in Section 05. Finish with Questions 16–19 from the worksheet. Listen for explanations based on data and tasks; a domain name without a reason is not yet evidence of understanding.

Worksheet

Name: Class & Section: Date:

A · Choose the best answer

1Which input most clearly belongs to a Computer Vision task?

(a) A table of monthly sales(b) Frames from a gate camera(c) A typed customer message(d) A list of bus fares

2A model labels a message as “request”, “complaint” or “praise”. Which technique describes its output?

(a) Classification(b) Regression(c) Clustering(d) Fixed automation

3Which job is mainly Natural Language Processing?

(a) Finding a helmet in a photograph(b) Estimating tomorrow's shop sales(c) Translating a spoken sentence(d) Grouping rainfall records

4A corridor light turns off five minutes after its motion sensor stops detecting movement. Its rule never changes. This task is best described as:

(a) AI prediction(b) Computer Vision(c) NLP(d) Automation

5A model estimates that a delivery will take 26 minutes. Which technique fits this numerical output?

(a) Classification(b) Regression(c) Clustering(d) Translation

B · Classify the domain

Write DS (Data Science), CV (Computer Vision) or NLP. Choose the main domain for the task as described.

Case	Domain	Reason: input and task
6. A library uses past borrowing rows to estimate which books will be in demand next month.
7. A kiosk reads a printed roll number from a photograph of an identity card.
8. A chat assistant identifies the purpose of a typed question.
9. A road camera locates each vehicle in an image.
10. A canteen uses a table of earlier sales to predict how many lunches to prepare.
11. A tool turns a Hindi voice recording into an English sentence.

C · Predictive techniques and data

12A parcel is assigned to “local”, “regional” or “national”. Name the predictive technique and explain why.

13A weather model estimates rainfall in millimetres. Name the predictive technique and explain why.

14A reading app receives unlabelled borrowing records and forms groups of students with similar reading choices. Name the technique and explain why.

15A team has 1,000 records. It uses 600 to teach a model, 200 while adjusting it, and keeps 200 unseen for the final check. Name the role of each portion.

D · Reason from the mechanism

16Why does adding a sensor to the corridor light in Question 4 not automatically make its task AI?

17Give one reason why a prediction based on past data may be wrong for a new case.

18A farm tool studies a photograph of a leaf and a table of soil moisture and weather values. It then suggests whether the plant needs attention. Name the two AI domains involved and state the job of each.

E · Think harder

19A school help-desk system reads a photograph of a handwritten Hindi complaint, identifies its meaning and mood, then estimates a number of hours for resolution from past case records. A student calls the whole system “only NLP”. Explain why that answer is incomplete. Name every domain involved, the input each uses, and the predictive technique used for the time estimate.

Answer key & teaching notes

Q	Answer	What to watch for
1	(b)	Frames are visual data. A camera alone does not prove AI; the task performed on its frames matters.
2	(a) Classification	The named categories are fixed before the prediction. Do not accept clustering.
3	(c)	Both speech and translation are language work. Some students choose (a) because it also uses AI, but it is vision.
4	(d) Automation	The sensor supplies an input; the unchanged five-minute rule decides the action.
5	(b) Regression	The output is a numerical value, not a named group.
6	DS	Input: organised borrowing records. Task: find a pattern and estimate future demand.
7	CV	The system begins with pixels and finds text in an image. Interpreting the words would add NLP, but that is not the stated task.
8	NLP	The input is written language and the task is to infer the writer's purpose.
9	CV	Locating objects in an image is a vision task, not Data Science merely because coordinates may be produced.
10	DS	Rows of past sales are compared to estimate a future quantity.
11	NLP	Speech recognition and translation both work with human language.
12	Classification	The answer must mention selection among predefined parcel categories.
13	Regression	Rainfall in millimetres is a numerical estimate. “Prediction” alone is too broad.
14	Clustering	The records arrive without group labels; similarity is used to form the groups.
15	600 training; 200 validation; 200 test	Validation supports adjustment during development. Test data stays aside for the final check.
16	It still follows a fixed “no motion for five minutes → off” rule and learns no pattern from examples.	“It has no camera” is not a valid reason; AI does not require a camera.
17	Any one: conditions changed; relevant cases were missing; records contained errors; or the new case differs from the training examples.	Do not accept “AI is always wrong”. The point is uncertainty, not uselessness.
18	CV examines the leaf photograph; DS compares the organised soil and weather values to support the estimate.	Students must connect each domain to its own input. NLP is not involved because no language task is stated.
19	CV reads writing from image pixels; NLP works with the Hindi words to infer meaning and mood; DS uses organised past-case records. Regression estimates the numerical resolution time.	This discriminates transfer from recall. Full credit needs all three input-to-task links and regression. Accept that optical text recognition hands extracted text from CV to NLP.

Mark the reasoning chain For Questions 12–19, award the explanation as well as the term. A correct label reached for the wrong reason will fail on a slightly different example. Ask: “What went in, what job was done, and what came out?”

Unplugged activity · “Domain Dispatch”

Period 8 · 35 minutes · No devices or internet · Designed for 40–48 students.

What you need

The classroom board, chalk, and one notebook page per team. Divide the class into eight teams of five or six. No printed card set is required.

Set up

Draw three large board columns headed Data Science, Computer Vision and NLP. Under them write the prompts input?, task? and output?. Each team copies a six-row answer grid and gives every member a role: reader, input finder, task finder, domain chooser, checker and reporter. In teams of five, the checker also reports.

Case bank

Past electricity readings are used to estimate next month's units.
A camera spots cracks in a road image.
A typed review is sorted as positive, negative or neutral.
A timer rings the school bell at the same times daily.
A photograph of a notice is read, then its text is translated.
Past bus trips are compared to estimate arrival time.

How it runs

Read one case aloud and leave it visible on the board. Teams get 90 seconds to record the input, task, output and domain. They may write automation or more than one domain when justified.
Call one reporter. Another team may challenge only by naming a different input-to-task link. Settle the case, then rotate all roles.
For Cases 1, 3 and 6, ask whether the output is classification or regression. For Case 5, draw an arrow from CV to NLP.
Score one point for the domain and one for the reason. The reason point prevents lucky guessing.

Board key 1: DS, regression. 2: CV, likely classification or object localisation depending on the output stated. 3: NLP, classification. 4: automation, no AI domain required. 5: CV reads the photographed text, then NLP translates it. 6: DS, regression.

Where the room usually goes wrong Teams may sort by object: “camera means CV” or “numbers mean Data Science”. Insist on the job. A camera used only to record video is not automatically doing vision analysis; numbers can be the output of any domain. With more than 48 students, add a ninth team and keep the same role rotation.

Project brief & rubric

Use one chart or up to three A4 pages. Students may work individually or in pairs. No working software, internet research or personal data is required.

Brief given to students “Design a Three-Domain School Service” — Choose one school need, such as library access, water use, bus arrival or lost property. Design three connected AI features: one using organised records, one using an image, and one using written or spoken language. For each feature show input → task → output. Identify one use of classification, regression or clustering. Add one fixed automation rule, clearly labelled as automation, and one warning about data quality, privacy or unfair results. This is a paper design, not a claim that the system has been built.

Criterion	4 — Exceeds	3 — Meets	2 — Approaching	1 — Beginning
Domain accuracy	All three domains are correct and connect logically	All three are correctly matched to tasks	Two are correct	One or none is correct
Input → task → output	Every chain is specific, complete and plausible	Three complete chains are shown	Chains have gaps or vague data	Mechanism is mostly missing
Prediction and limits	Technique, automation and warning are all accurate and well explained	All three are present and accurate	One element is missing or confused	Two or more are missing
Communication	Connections are exceptionally easy to follow and choices are defended	Work is organised and readable	Meaning can be followed with effort	Work is incomplete or unclear

Suggested score: 16 marks. Record both the total and a one-sentence note about the student's reasoning; the note is more useful than a percentage alone.

Evidence record

Complete one record for each class and section. Attach or file the named samples with it.

Field	Record
School
Class & section
Chapter taught	AI Ch. 1 — AI Domains and Applications
Dates
Periods used
Teacher
Activity conducted	Domain Dispatch (unplugged domain classification)
Assessment used	Three-Domain School Service project, rubric-scored
Students assessed
Class outcome	Secure: ____ Developing: ____ Needs follow-up: ____
Common misconception noticed
Samples retained	☐ 3 marked projects ☐ 3 worksheets ☐ Activity record, or a photograph of the work (no children or names visible) if school policy permits
Teacher's next step
Teacher signature & date

Minimum useful evidence Keep this signed sheet, three rubric-marked projects and three completed worksheets. If school policy permits a photograph, photograph the work rather than the children, keep student names out of the frame, and follow the school’s consent procedure. Evidence stays in the school; never send it to ai4kiddos.in.