What is overfitting — when ai memorises instead of learns — explained for kids?

Why an AI that aces practice questions can still fail in the real world.

What's the most common mistake children make about this concept?

100% training accuracy means a great model — it is actually a red flag for overfitting in most cases.

How does Dhee Learning teach this in a Class 6 session?

Dhee opens with a question — for example: "A rickshaw-fare app scores perfectly on its 10 practice rides but fails everywhere new. What happened inside it, and what is that called?" — 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.

Overfitting — when AI memorises instead of learns

What this concept actually says

Overfitting is when a model learns the training data too perfectly, including its noise and quirks
An overfit model scores high on training data but poorly on new, unseen data
Overfitting is like memorising answers without understanding the subject

An analogy your child will recognise

Guidebook mugging before an exam

Rohan memorises the guidebook's 40 solved questions word for word. On those exact 40 he scores full marks. The real exam asks fresh questions from the same topics — and he fails. He never learned the method; he stored the answers. FaceFind did exactly this with its 40 practice photos.

One-route auto driver

An auto driver who drives one single route every day memorises every pothole on it. Put him on a new road and the memorised potholes are useless — he never learned how to read roads in general. And driving that same route MORE only makes him better at that one route.

Common misconceptions to watch for

100% training accuracy means a great model — it is actually a red flag for overfitting in most cases.
Overfitting is rare in practice — it is one of the most common problems practitioners face, especially with small or narrow datasets.

Key facts in one breath

The gap between training accuracy and test accuracy is the clearest signal of overfitting — the larger the gap, the worse the overfit.
Overfitting is more likely with small datasets, very complex models, or very long training.
Techniques to fight overfitting include: using more data, adding dropout (randomly switching off neurons during training), and early stopping.
Regularisation is a mathematical technique that penalises overly complex models during training to discourage memorisation.

How Dhee Learning teaches this — the 3-stage question loop

Every Dhee Learning 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

A rickshaw-fare app scores perfectly on its 10 practice rides but fails everywhere new. What happened inside it, and what is that called?

Rote answer

"Overfitting is when the model learns too much from training data."

Understood

"The app stored its 10 exact practice rides — routes, timings, accidents — instead of learning how fares actually work. New rides find nothing to match, so it fails. Perfect-on-practice plus lost-on-new is the overfitting signature: memorised examples, not a learned pattern."

Stage 2 — Reasoning

One builder wants many varied rides; another wants ten times longer training on the same 10 rides. Why does only one of these fix the failing fare app?

Follow-up Dhee may use: If memorising too much causes this, what would the opposite failure — a model too simple to learn even its practice rides — look like in the two scores?

Stage 3 — Application

Your cousin memorised last year's paper, scored 100 on it, and the real exam is tomorrow. Using the overfitting idea, what do you advise tonight — and how would they prove it worked?

Misconception Dhee watches for: Thinking overfitting only happens when accuracy is low — an overfit model has very high training accuracy; the problem shows up on new data.