Guide

What is an AI class for kids, and what happens in one?

By Viraj Samarasinghe & Menura Dulkith · Updated

An AI class for kids teaches children how artificial intelligence works by having them build with it, instead of just hearing about it. They train a model on their own examples, watch it fail, and work out why. A good one runs live, has the child make something in the first month, and pairs the AI with real electronics.

The short version

A children's AI class is a hands-on workshop. A child builds small things that make decisions, and learns the ideas by watching those things work and fail. It isn't a set of lectures about artificial intelligence.

Be careful here, because both kinds are sold under the same name. If a child mostly watches demonstrations of AI tools, that's a computer-literacy class. If a child trains something, breaks it and fixes it, that's an AI class. One question tells you which one you're looking at: what will my child make in the first month?

What a typical week looks like

A live session of about ninety minutes, with a group small enough that a child can ask a question during it. The instructor spends some of that time introducing an idea. Most of it is the children working, while the instructor spots the particular wrong idea each one has picked up. A recorded video can't do that, and it's why live classes matter more at this age than at any other.

Between sessions there's usually something to finish or extend. It isn't homework in the school sense. More often it's a thing that half works, and the child wants to make it work.

Three things a good class helps a child understand

First, a model learns from examples. A normal program is told what to do. A model is shown examples and works out a rule for itself. Children get this by training a classifier on thirty photos, then showing it a thirty-first it has never seen, and watching it be confidently wrong.

Second, the training data decides everything. After that failure, the obvious question is why. The answer is that it only ever saw examples like the ones you gave it, and that idea is the base of everything a citizen needs to know about AI. It's also where 'bias' becomes something real instead of an abstract word.

Third, a confident answer can still be wrong. Models give a confidence number, and children soon notice that a wrong answer can come with a high one. For a generation that will use these systems every day, this may be the most useful thing in the whole syllabus. It also applies straight away to the chatbots they already use.

Why the electronics matter

Most good classes at this age pair the AI with hardware: a microcontroller such as an Arduino, some sensors and motors. It isn't there to make the brochure look nice. A sensor gives you a number, the number is noisy, and deciding what to do with it is the real engineering.

It's also where children learn to keep going. Wiring is unforgiving, and a loose connection doesn't care how well you understood the theory. Children who finish a term of this get noticeably better at debugging everything else.

What it should cost, and what to check

Look at the total cost of the whole programme, including any hardware kit, rather than the monthly fee. In Sri Lanka the kit can quietly end up costing more than the tuition. Ask for the total in writing, and ask whether your child keeps the kit.

Ask who teaches the sessions and what they studied. You can check that, but you can't check “expert instructors”. Ask whether the sessions are live. And ask what happens when a child's project doesn't work. A class with a good answer to that is teaching engineering. A class without one is just running demonstrations.

Who it suits

Children from about 9, when most can hold a variable in mind and see a wrong output as useful information instead of a failure. The ones who do best often aren't the children who are already on a screen all day. They're the ones who build things and take things apart.

If reading instructions on a screen is still the hard part, or a wrong result always ends the session, waiting a year costs nothing. Starting early and having it go badly can make a child believe this isn't for them, and that's much harder to fix.

Related questions

How is an AI class different from a coding class?

A coding class teaches a child to tell a computer exactly what to do. An AI class teaches them to build something that works out what to do from examples. AI classes include programming (you can't avoid it), but the main idea is different: the system is trained from examples instead of given instructions. That idea is what the class is really about.

What age is an AI class for?

Most programmes aimed at children run from about 9 to 14. Under 9, logic and building are the more useful groundwork. Over 14, a student can usually move faster and start on the mathematics if they want it.

Can an AI class be taught online?

Yes, as long as the sessions are really live and the group is small enough for a child to ask questions. In our experience, recorded-video courses work less well for under-14s, because most of the teaching happens when someone catches a child's wrong idea in the moment.

Does my child need a powerful computer?

No. Beginner AI work for this age group runs on an ordinary laptop or desktop, and the training involved is small. What you do need is a laptop or desktop instead of a phone, because the tools are awkward to use on a phone.

See a real session before deciding

kidslab.lk runs a free introductory seminar for children aged 9–14, live online across Sri Lanka. No obligation to enrol — it exists so you can answer most of the questions above for yourself. Where children join from.

Register for the free seminar →

More guides