About halfway through the second lesson of our AI & Python course, something reliably happens. A student runs the text generator they've just written, trained on the lyrics of a song they love, and it produces a line so confidently ridiculous that the whole room cracks up. Then someone else runs theirs. Then everyone wants to know why their version said something different.
That moment is the course working exactly as designed. The laughing is real fun. The "wait, why did it do that?" is the start of some genuinely hard thinking. We built the course to have plenty of both.
Quick Answer: Our AI & Python course is meant to be fun, and it's meant to be a proper mental workout. Kids and teens build their own language model in Python, the idea behind ChatGPT, and spend the course making it smarter, stranger and funnier. Along the way they learn how AI really works, which helps them navigate a world full of it, and they learn to read and write real code. That second part matters more than ever: a child who uses AI without understanding code might get 10x more done, but one who understands code can direct AI, catch its mistakes and get something closer to 100x.
Fun Is the Point, Not a Bonus
A lot of courses treat fun as a sweetener: get through the serious material, then reward the class with a game at the end. We think that's backwards. When kids are enjoying themselves, they ask more questions, try more things, stick with problems for longer and come back next week wanting more. Fun isn't a break from learning. For most children, it's the fuel for it.
Fun doesn't mean easy, though. The most fun a child has in a lesson is usually right after they've cracked something difficult. A game that's too easy gets boring in minutes; a game that's hard but fair keeps them playing for hours. We try to make every lesson feel like the second kind.
What Fun Looks Like in an AI & Python Class
The fun in this course comes from the AI itself, not from activities bolted on around it. A few of the moments students tend to remember:
- Lesson one's glorious nonsense. Their very first program strings words into "sentences" that make no sense at all. It's meant to. Making it less terrible becomes the mission for the rest of the course.
- Training it on things they love. A favourite book, a film script, song lyrics, a classic novel from Project Gutenberg. Students pick the text, and their model starts sounding a bit like it, in ways that are often hilarious.
- The temperature dial. One setting controls how safe or wild the output is. Turn it down and the model plays it boring. Turn it up and it goes off the rails. Every student turns it all the way up at least once, just to see.
- The strawberry mystery. Why do real AI chatbots sometimes miscount the letters in "strawberry"? Students find out by rebuilding their model to work on pieces of words, and the answer is far more satisfying than being told.
- Catching the big models out. Using DotCode AI, students give the same prompt to different frontier models and compare what comes back, including the confident mistakes.
- Demo day. The course ends with each student showing off a project of their own, a lyric generator, a story machine, a bot that writes like their favourite author, and explaining what it gets right, what it gets hilariously wrong, and why.
If you'd like a closer look at the projects, we've written up what our students actually build in AI classes, and there are more ideas to try at home in our list of AI project ideas for teens.
A Mental Workout, Too
Underneath the jokes, every lesson asks students to think hard. Not memorise, think. Why does always picking the most likely next word make the program get stuck in a loop? Why does looking back three words make the writing better, and then suddenly make it copy the original text word for word? How can a program work out that "king" and "queen" are related when nobody told it?
Questions like these exercise exactly the kind of thinking that's hard to train any other way:
- Logical reasoning. Code does precisely what you tell it, so students learn to think in clear, careful steps.
- Breaking problems down. "Make it write better" is impossible. "Make it look at the previous two words instead of one" is something they can do this afternoon.
- Debugging. When the program misbehaves, students become detectives: form a theory, test it, rule it out, try the next one. It's frustrating for a few minutes and hugely satisfying when they find it.
- Probability and maths in action. Weighted random choices and turning words into numbers make abstract maths suddenly feel useful.
- Persistence. Nothing works first time. Students get used to that, and get good at pushing through it.
Think of it like a gym session for the brain. It's tiring in a good way, and students leave sharper than they came in. That kind of real understanding is also something AI can't fake on a child's behalf, as we explain in why AI can fake the homework but not the explanation.
Understanding AI Helps Kids Navigate the World
Today's kids are growing up with AI everywhere: chatbots that help with homework, apps that recommend what they watch next, images and videos that may or may not be real. Most of them use these tools confidently. Very few understand what's going on inside.
Once a student has built a language model themselves, that changes. They know a chatbot is predicting a likely answer rather than looking up the truth. They know there's randomness built in, which is why it can answer the same question two different ways. They know it can sound completely sure and still be wrong. That understanding makes them better users of AI and harder to fool by it.
We go into how this works in how ChatGPT actually works, and into which tasks AI is good at, and which need checking, in learning with AI without depending on it. For a broader starting point, our parent's guide to AI for kids covers the basics.
Understanding Code: Not 10x, but 100x
AI can write code now, which leads some parents to wonder whether learning to code still matters. It matters more, because AI is a multiplier, and a multiplier depends on what it's multiplying.
Picture two teenagers using the same AI tool to build a small game.
The first can't read code. They describe what they want, copy what the AI gives them and run it. When it works, that's brilliant, maybe 10x more than they could have done on their own. But the moment something breaks, they're stuck. They can only paste the error back in and hope, and often the AI goes round in circles.
The second has learned Python. They read what the AI wrote, spot the bug, and tell it exactly what to fix. They know what to ask for in the first place, they can tell good code from code that only looks good, and they can take the project much further than the first student ever could. That's where 100x comes from.
Our AI & Python course teaches real Python from scratch, lists, loops, dictionaries, functions and files, because students need every one of those to build their model. By the end, code isn't a mysterious wall of text the AI produces. It's something they can read, question and improve. We make the full argument, including why the gap grows even wider for professional developers, in is there any point learning to code now that AI can code? If you're wondering where a teen should begin, see Python for AI: how a teenager should actually start.
Why Fun and Hard Work Belong Together
The fun and the mental workout aren't two separate things we've balanced against each other. They feed each other. The fun is what keeps students pushing through the hard parts, because they really want to see what their model does next. And the hard parts are what make the fun satisfying, because the funny output on screen came from code they wrote and understand.
That's the experience we want every student to have: laughing in one minute, frowning at a bug the next, and leaving with skills that will make them far more capable with AI for years to come.
Is the AI & Python Course Right for Your Child?
Our AI & Python course is for teens aged 13–17 and needs no coding experience. It runs over 10 lessons of 1.5 hours, in small groups of up to 6, in person at our Jurong East studio or live online. Every student gets access to DotCode AI throughout. If your teen would rather do it in one go, there's also an AI & Python holiday camp that runs Monday to Sunday.
Students who finish can carry on to AI Image & Video Generation, where they turn their text generator into a picture generator. Younger kids can start with Roblox or game development and come to AI with some code behind them.
Every course starts with a free 45-minute trial lesson. Book a free trial class and let your child see how much fun a hard problem can be.
