Class 5 · CBSE AI · Strand B — Prediction & Probability
How does YouTube recommend videos? Recommendation engines for kids
The hidden AI that decides what your child sees next — and why it matters.
What this concept actually says
- Recommendation engines predict what you will want next, based on your past behaviour and the behaviour of similar users
- They use two main approaches: what you liked before (content-based), and what people like you liked (collaborative filtering)
- Recommendation engines are designed to maximise engagement, not necessarily your wellbeing
An analogy your child will recognise
Neighbourhood bookshop
A good bookshop owner who knows you well says: 'You loved that mystery novel — you'd probably like this one too.' That's a content-based recommendation. Now imagine she's tracked what a hundred customers with your exact taste bought next — that's collaborative filtering. Netflix does both, at the scale of millions.
Sabzi mandi vendor
A vegetable vendor who knows every regular customer remembers: 'She always buys spinach when she buys paneer — let me suggest spinach.' That's pattern-based recommendation. Amazon's 'Customers who bought this also bought...' is the same thing, scaled to a billion users.
Common misconceptions to watch for
- YouTube and Netflix recommend things because they think those things are good for you.
- Recommendation engines show you a balanced or random sample of available content.
Key facts in one breath
- Recommendation engines use two main methods: content-based filtering (based on your own history) and collaborative filtering (based on similar users' behaviour).
- Most major platforms optimise recommendations for engagement (clicks, watch time) rather than user satisfaction or wellbeing.
- Recommendation engines are prediction machines: they predict which item you are most likely to engage with next.
- The data used to train a recommendation engine determines whose preferences get amplified and whose get ignored.
How Dhee teaches this — the 3-stage Socratic loop
Every Dhee session for this concept follows three stages. We share the questions Dhee actually asks, so you can hear what a session sounds like.
Stage 1 — Surface
When YouTube suggests a video after you finish watching one, how do you think it decides what to show? Is it random?
Rote answer
"It suggests videos based on what you watched before."
Understood
"It looks at what I've watched and liked, but also at what millions of other people with similar taste watched after that same video — and recommends what kept them watching longest. It's not just about me; it's about a pattern across many people."
Stage 2 — Reasoning
Netflix wants you to watch as much as possible. YouTube wants you to watch as long as possible. So their recommendation engines are designed to maximise watch time. How might that goal affect what they recommend to you?
Follow-up Dhee may use: Could a recommendation engine ever recommend something that's good for you but that you wouldn't click on immediately? What would that look like?
Stage 3 — Application
Design a recommendation engine for a library — physical books. What data would you use to predict what a child wants to read next? How is this different from what YouTube does?
Misconception Dhee watches for: Child assumes recommendation engines exist to help you find what you want, without recognising the platform's separate engagement-maximisation goal.
Want your child to actually understand this?
Spark turns this concept into a 15-minute spoken session — asking, listening, and probing — so your child builds the idea themselves.
Frequently asked questions
What is recommendation engines — explained for kids? +
The hidden AI that decides what your child sees next — and why it matters.
What's the most common mistake children make about this concept? +
YouTube and Netflix recommend things because they think those things are good for you.
How does Dhee teach this in a Class 5 session? +
Dhee opens with a question — for example: "When YouTube suggests a video after you finish watching one, how do you think it decides what to show? Is it random?" — listens to your child's answer, then probes the reasoning behind it. The session ends when the child can apply the idea to a brand-new situation, not just recall it.