CBSE’s AI Curriculum for Classes 3–8: What It Actually Covers, Grade by Grade
Your Class 3 child is not learning AI. CBSE’s new Computational Thinking and Artificial Intelligence curriculum, rolling out from the 2026–27 session, gives Classes 3–5 fifty hours a year of puzzles and pattern work folded into maths. There is no AI in it. AI itself begins in Class 6, at twenty hours a year. Here is the whole thing, unit by unit, from the official document.
Most of what you’ll find written about this is wrong in the same direction: it reads “AI from Class 3” and imagines eight-year-olds training neural networks. The actual curriculum is more modest, better designed, and considerably more interesting than the panic suggests.
The one-line version
| Classes 3–5 | Classes 6–8 | |
|---|---|---|
| Hours per year | 50 | 100 |
| AI content | None | 20 hours |
| Delivered as | Embedded in Maths and The World Around Us | Specific AI topics + interdisciplinary projects |
| New subject? | No | No |
| Board exam? | No | No |
| Graded by | Written CT questions, group activities, Teacher Observation Journal | Practicals, thematic projects, reflections, discussions |
CBSE is explicit that the hours are “suggested to maintain the existing curriculum load” - schools can go further, but nothing is being added to your child’s bag.
Classes 3–5: fifty hours, and not a single hour of AI
For the preparatory stage, computational thinking is not taught as a topic. It is threaded into subjects your child already studies - Mathematics and The World Around Us - through a resource book that mirrors the textbook’s table of contents chapter for chapter, adding CT questions to each one.
So the delivery mechanism is: same maths chapter, extra questions of a particular shape.
Four skills are being built, and they progress each year:
Abstract thinking. Class 3 works with different viewpoints of 3-D objects, shapes after flips and turns, and missing parts of incomplete patterns. By Class 5 this has become multi-layered reasoning.
Pattern recognition. Class 3 identifies simple patterns with one or two changes between terms - numbers, shapes, letters or a mix. Class 4 handles multi-rule patterns; Class 5, complex progressions.
Decomposition. Class 3 breaks down problems with two or three clues: number clues, the faces and edges of 3-D objects, step-by-step exchanges of money or objects, tables holding several pieces of information at once. Class 5 handles interconnected problems.
Algorithmic thinking. Step-by-step rules applied to number sequences, movement on grids, events ordered by before/after/in-between clues, and multi-step instructions involving moves and transfers.
If that sounds like a good maths olympiad paper, that’s roughly right. It is the reasoning layer that sits under arithmetic, taught deliberately rather than by accident.
Classes 6–8: where the hundred hours actually go
This is where AI arrives, and it is a fifth of the time.
| Component | Hours/year | What it is |
|---|---|---|
| Advanced CT skills | 40 | Builds on Classes 3–5, applied to complex situations |
| Introductory AI concepts | 20 | Foundational AI; exposure to AI tools |
| Interdisciplinary projects | 40 | Cross-subject projects applying CT and AI |
The largest single block is not AI. It is projects - forty hours a year of work that deliberately cuts across Mathematics, Science, Social Studies and English. CBSE describes these as “curated to make learning engaging and hands-on rather than academically demanding.”
That phrasing is doing real work. It tells you the board expects this to be assessed on whether a child can apply something, not whether they can recall it.
The Early AI syllabus, unit by unit
Twenty hours per class, four units of five hours each. This is the part nobody publishes, so here it is in full.
Class 6 - twenty hours
| # | Unit | Hours |
|---|---|---|
| 1 | Introduction to AI and everyday examples - what AI is; AI vs automation; human vs machine intelligence; types of learning (supervised, unsupervised, reinforcement) | 5 |
| 2 | Basic data concepts - data types (numbers, text, images, sound); organising and representing data in tables and charts | 5 |
| 3 | Simple pattern recognition and decision making - patterns in data and daily routines; decisions from observation | 5 |
| 4 | Ethics and digital responsibility - online safety, privacy, passwords, digital footprints | 5 |
Class 7 - twenty hours
| # | Unit | Hours |
|---|---|---|
| 1 | AI domains - classification, regression and clustering with hands-on practice on a small dataset; computer vision, NLP, data science; chatbots, image recognition, translation | 5 |
| 2 | AI in industries - healthcare, education, transport, communication | 5 |
| 3 | Data visualisation and analysis - collecting structured data; bar, line and pie charts; interpreting patterns | 5 |
| 4 | Ethics and AI bias awareness - bias in AI, case examples, fair use, digital citizenship | 5 |
Class 8 - twenty hours
| # | Unit | Hours |
|---|---|---|
| 1 | AI project lifecycle - define problem, collect data, test AI tools, reflect and improve | 5 |
| 2 | Deeper dive into applications - environment, healthcare, automation, education; hands-on with no-code tools (image classifiers, chatbots, prediction apps) | 5 |
| 3 | Data and fairness - how AI uses data; identifying bias in datasets; strategies for fairness | 5 |
| 4 | Ethics and responsible AI - privacy, misinformation, social impact; reflection on real-world challenges | 5 |
Notice what recurs. Every single year, one of the four units is ethics. A quarter of all AI teaching time from Class 6 to Class 8 is about bias, privacy, misinformation and responsible use. That is a deliberate curricular choice and, given what your child is already doing with chatbots outside school, a defensible one.
There is no AI exam. Here’s how it’s assessed instead.
No separate subject, no board paper, no marks out of a hundred. CBSE lists the methods by stage:
Classes 3–5 - written tests containing CT questions and puzzles; interactive group activities such as treasure hunts; a Teacher Observation Journal.
Classes 6–8 - written tests; interactive group activities; practical examinations; a Teacher Observation Journal; thematic projects; reflections and group discussions.
The curriculum says assessment should target “the ability to apply knowledge... creativity, and the development of ethical reasoning skills, rather than assessing rote memorisation.”
Which means the real rubric, whatever the school writes down, is this: can your child explain what they did and why? A project that was quietly finished by a parent the night before produces a child who cannot answer a follow-up question, and teachers notice.
What this asks of parents
Very little, formally. Nothing is being added to the timetable and there is no exam to coach for. Schools may also choose their own tools - CBSE recommends free and open-source options and explicitly declines to tie schools to any particular software, so you should not be buying anything.
The thing worth knowing is where the implementation risk sits. Roughly 8 to 8.5 million teachers across India need training for this, and by one 2026 estimate only about 15% are currently AI-fluent. A JanAI survey found 72% of government-school teachers supportive of AI in classrooms, but consistently short on training and institutional backing. Enthusiasm is not the bottleneck. Capacity is.
So the honest advice is narrow:
- Don’t buy an AI course for a Class 3 child on the strength of this. There is no AI in Classes 3–5. Anyone selling you one is selling against a syllabus they haven’t read.
- Ask the school one question at the next PTM - who is teaching this, and have they been trained? The answer tells you whether your child’s fifty or hundred hours will be real.
- When the Class 8 project comes home, don’t do it. Ask what problem they chose and why. The curriculum is built to reward the child who can answer that.
Frequently asked questions
Is AI compulsory in CBSE schools now?
Computational Thinking and AI are mandatory for Classes 3–8 from the 2026–27 session, but not as a separate subject. CT is embedded in existing subjects; AI appears as specific topics from Class 6 onward.
Does my Class 4 child have to study artificial intelligence?
No. Classes 3–5 cover computational thinking only - abstract thinking, pattern recognition, decomposition and algorithmic thinking - integrated into Mathematics and The World Around Us.
Will there be a board exam for AI?
No. Assessment for Classes 3–8 is through written CT questions, practical work, projects, group activities and a Teacher Observation Journal.
How many hours a year is it?
Fifty hours a year for Classes 3–5 and a hundred for Classes 6–8. For Classes 6–8 that splits into 40 hours of advanced CT, 20 of AI, and 40 of interdisciplinary projects.
Do we need to buy software or a laptop for this?
CBSE recommends free and open-source tools and gives schools flexibility in what they use. Nothing in the curriculum requires a personal purchase.
What is The World Around Us?
TWAU is the integrated environmental-studies subject at the preparatory stage under the new curriculum framework. Along with Mathematics, it is where computational thinking is embedded for Classes 3–5.
Related reading
- 86% of Kids Aged 9–17 Now Use AI - the data on how children are already using AI outside the classroom.
- The Protégé Effect - why a child explaining a topic back teaches them more than being taught it.