CBSE AI Syllabus Class 10 - What’s Actually in It

Ask any Class 10 student what the AI paper is about and you’ll probably get a shrug. Ask their teacher and you might get a rehearsed answer about “future-ready skills.” Ask the school principal and — nine times out of ten — they’ll tell you it’s a scoring subject.

None of these answers are entirely wrong. But none of them are quite right either.

The CBSE AI syllabus for Class 10 — Subject Code 417— is genuinely one of the more thoughtfully designed additions to the school curriculum in recent years. It’s just that most students never experience it that way, because somewhere between the NCERT framework and the classroom, the intention gets lost.

So let’s go through what it actually says. Unit by unit, mark by mark — and more importantly, why it’s designed the way it is.

First, the Structure (Because the Marks Distribution Says a Lot)

Total: 100 marks. Split right down the middle — 50 theory, 50 practical.

That 50-50 split is deliberate, and it’s worth pausing on. In most school subjects, practicals are an afterthought — a few diagrams, a lab report, some viva questions you can mug up the night before. Not here. Half the assessment requires students to actually do something: write Python, process data, run image operations, present a project. You can’t fake your way through that with good handwriting and a good memory.

The theory side is divided into Part A (Employability Skills, 10 marks) and Part B (Subject-Specific Skills, 40 marks). Part A covers communication, self-management, ICT, entrepreneurship, and green skills. Sounds like filler — it’s not. These are the competencies that separate someone who can work in an AI environment from someone who can only describe one.

Part B is where things get interesting.

The Six Units — And What They’re Really Teaching

Unit 1: Introduction to AI — Smarter Than It Sounds

Most intro chapters in school textbooks are forgettable. This one has a shot at being different, if it’s taught right.

The unit opens with a question that sounds simple but isn’t: what is intelligence? Not artificial intelligence — just intelligence. How do humans make decisions? What does reasoning actually look like? Students work through this before they ever touch a definition of AI, which is the correct order. You can’t understand what a machine is approximating until you understand what it’s approximating.

From there, it moves into the three domains the whole course is built on: Data Science, Computer Vision, and Natural Language Processing. Each one is introduced through a real, interactive tool rather than a definition. Data Science through an impact filter that shows how rising temperatures affect ecosystems. Computer Vision through AutoDraw. NLP through Wordtune.

And then — this is the part most people don’t expect from a Class 10 syllabus — it covers AI ethics. Not as a box-ticking exercise but through MIT’s Moral Machine, where students confront actual trolley-problem scenarios involving self-driving cars. The questions it raises don’t have clean answers. That’s the point.

If a 15-year-old walks out of this unit understanding that AI isn’t neutral — that it reflects the choices of the people who build it — that’s a genuinely valuable thing to know.

Unit 2: AI Project Cycle — The Unit That Should Be the Whole Course

If there’s one unit in this syllabus that has the potential to change how a student thinks, it’s this one.

The AI Project Cycle — problem scoping, data acquisition, data exploration, modelling, evaluation — is essentially the same framework used by data scientists at actual companies. Not a simplified version. The real thing, introduced at an age-appropriate depth.

The problem scoping section is what sets this apart. Students aren’t just asked to pick a project topic — they’re asked to connect it to the UN’s Sustainable Development Goals. Why does this problem matter? Who does it affect? What would a solution actually change? That’s a level of critical framing that most undergraduate courses don’t bother with.

In practice, this unit often gets rushed — because it’s easier to teach definitions than process. That’s a mistake. Students who genuinely understand the project cycle leave Class 10 knowing how to think about problems, not just how to answer questions about them.

Unit 3: Advanced Python — Practical Only, But Don’t Underestimate It

Python shows up only in the practicals — no theory marks attached — which leads some students to deprioritise it. That’s backwards.

Everything else in the syllabus that involves actually doing something — data analysis, image processing, text handling — runs through Python. Jupyter Notebooks, variables, control structures, built-in functions: these aren’t standalone concepts. They’re the plumbing. Get shaky here and Units 4, 5, and 6 become a struggle. Get solid here and suddenly the rest of the course starts making sense in a way it doesn’t from just reading about it.

Unit 4: Data Science — Where Numbers Start Making Sense

NumPy, Pandas, Matplotlib. Mean, median, mode, standard deviation. On paper this looks like a stats class. In practice it’s something different — because students aren’t just computing these values, they’re asking why.

Why visualise data before modelling it? Because patterns you can’t see in a table become obvious in a scatter plot. Why does standard deviation matter? Because an average without spread is nearly meaningless. The curriculum threads these questions through the technical content in a way that, again, many first-year college courses don’t. Students who actually engage with this unit — rather than memorising library syntax — come out with a statistical intuition that’s surprisingly hard to teach later.

Unit 5: Computer Vision — The One Students Actually Enjoy

This unit tends to land well with students because the applications are immediately obvious. Face unlock on your phone. Tumour detection in radiology. Quality control on a factory floor. Every example connects to something real.

The technical content covers pixels, RGB values, grayscale images, and eventually OpenCV — Python’s computer vision library — for hands-on image processing. There are good activity hooks too: the Emoji Scavenger Hunt for object recognition, a pixel art creator, an image kernel visualiser.

Convolutional Neural Networks are listed as optional/enrichment content. Don’t skip them entirely if your students are curious. CNNs are arguably the most consequential architecture in modern AI, and even a surface-level introduction at this stage creates an anchor that pays dividends later.

Unit 6: Natural Language Processing — The Hardest Unit to Teach Well

NLP is harder to make tangible than Computer Vision, because language is something students think they already understand. The challenge is getting them to see it the way a machine does — as a problem, not a given.

The syllabus covers chatbots, sentiment analysis, text normalisation, and the Bag-of-Words model — plus a hands-on pipeline where students actually clean and process text in Python. The Google Translate activity (using it on words with identical spellings but different meanings) is a clever entry point into ambiguity, which is the central problem of NLP.

A student who finishes this unit understanding why language is genuinely difficult for machines — not just technically but conceptually — is a student who’s thinking about AI at the right level of depth.

Unit 7: Evaluation — The Most Underrated Unit in the Entire Course

Ask most people how to judge whether an AI model is good and they’ll say: accuracy. High accuracy, good model. That’s wrong — and this unit explains why.

Imagine a model designed to flag patients at risk of a rare disease. If only 2% of patients actually have it, a model that classifies everyone as healthy achieves 98% accuracy. It’s also completely useless. Precision and recall exist precisely because accuracy lies in cases like this. The F1 score exists because you often need to balance them.

The confusion matrix activity — building one for a containment zone prediction model — forces students to sit with this tension rather than just memorise the formula. That’s good pedagogy. It’s also genuinely sophisticated thinking for any age, let alone Class 10.

The Practical Marks: Where the Real Learning Happens (or Doesn’t)

Fifty marks. Fifteen minimum programs. A practical exam covering Python, Data Science, and Computer Vision. A viva. A project or field visit.

On paper, this is the best-designed part of the whole assessment. In reality, it’s the part that varies most wildly across schools. In some classrooms, students write and debug actual Python code, present real projects, get asked real questions in the viva. In others, the “practical file” is copied from a senior’s notebook and the viva is two minutes of nodding.

The difference in what students carry forward from those two experiences is enormous. Not just for the exam — for everything after it.

Why This Subject Has More Long-Term Value Than It Gets Credit For

NEP 2020 talks a lot about skill-based learning. The CBSE AI syllabus for Class 10 is one of the few places in the curriculum where that phrase actually means something.

A student who’s genuinely worked through this syllabus — not just cleared the exam, but actually engaged with it — arrives at Class 11, at college, at their first internship, with things most of their peers don’t have. A working mental model of what AI is and isn’t. Practical Python exposure. The ability to frame a problem before trying to solve it. An instinct for asking “what does this metric actually mean” before trusting a number.

Those aren’t small things. They compound. The students who take this subject seriously at 15 are the ones who feel significantly less lost at 20 when everyone around them is scrambling to learn the basics of data and AI for the first time.

AI for Schools works with 250+ partner schools across India to deliver offline, in-school AI education — project-based, practically grounded, and mentored by experts who actually work in the field. As a Google Professional Development Partner, our programmes are built to go beyond what any textbook covers, while staying fully aligned with the CBSE AI syllabus and NEP 2020.

If your school wants students who genuinely understand AI — not just students who passed the paper — let’s talk.