At 13, She Started With Scratch. Five Weeks Later She Published a Guide to AI.
Yunwen Eric
··
7 min read
Iwu Ariel-Lamp is thirteen years old. She came to the Togeva STEAM Summer School this year to learn the basics — what an algorithm is, what a tag does, how to make a sprite move across a screen.
Five weeks later she had published Understanding AI: a fourteen-section explainer on artificial intelligence, written for students her own age and for the parents who worry about them. It is live on the internet. Its source code is open. She typed it by hand in HTML and CSS, with no framework, no template and not a single line of JavaScript.
We are proud of her. This post is about what she made, and why a thirteen-year-old writing this particular document matters more than it might sound.
Understanding AI opens with a definition many adults would struggle to give from memory: machines that can learn, notice patterns, and help people solve problems. Then it asks the question the whole site is built to answer — AI is already in how we learn, communicate, work and create, but what exactly is it, and how does it actually work?
Fourteen sections follow, and they move in three deliberate steps.
First, what it is. What separates a program that follows rules from a system that learns patterns. A section called "You Probably Used AI Today" that walks through the phone in your pocket — the maps, the recommendations, the spam filter, the translation. A timeline running from the 1950s to now.
Second, how it works. This is where the site is at its best, because she refused to hand-wave. To explain training, she uses mangoes: show a system enough of them and it learns what makes a mango a mango. Then she runs a worked example — "Let's Train a Fruit AI" — in which a model trained on three thousand images does not return an answer, but a confidence score. A guess with a number attached to it. For a reader of any age, that one idea is the difference between understanding AI and being impressed by it.
Third, what to do about it. Eight fields where AI is already being used. Seven reasons people reach for it. Eight ways a student can actually use it.
There is a version of this project where a clever teenager builds a nice-looking page about a fashionable subject. That is not what she made.
The heart of the site is a section titled "Use AI To Learn, Not To Copy." It sets good use against bad use in plain terms: ask it to explain, to quiz you, to check your reasoning, to help you sharpen your own words — do not hand it your assignment, and do not take it into a test. It tells readers not to give it their passwords, their address or their personal details.
And it carries one line that is better advice than most of what adults have written about AI this year: never confuse confidence with correctness.
She follows it with a section on asking better questions, reduced to a formula a twelve-year-old can hold in their head — what I want, plus context, plus how I want it.
That is not a website about AI. That is a teaching document.
The first is a section called "Explain AI At Home" — a script for a young person to sit down with a parent who has heard of AI, is suspicious of it, and has never once seen it work. It comes with five prompts to demonstrate, covering learning, language, creativity, school and code. She built her classmates a way to carry the summer school home.
The second is the challenge that closes the site: five days of small tasks, in which day four asks the reader to catch the AI making a mistake, and day five asks them to teach someone else what they have learned.
Catch it being wrong. Then go and teach it. From a thirteen-year-old, five weeks into her first line of code.
When UNESCO surveyed schools and universities in May 2023, fewer than 10% of them had any institutional policy or formal guidance on generative AI — 7% of schools, 13% of universities. Three years on, the technology has moved a great deal faster than the guidance has.
The gap is not theoretical. Students are already using these tools — for homework, for revision, and for things they should not be using them for — and in most classrooms nobody has sat them down and explained what the tool is, where it fails, and what using it well actually looks like.
Lamp's site is one answer to that gap, written from inside the group that needs it. Near the end she asks the question directly: what could AI mean for Africa? Not as a slogan, but as an open question about local problems, and about who will be equipped to solve them.
None of it started with artificial intelligence. It started with a sprite.
Her first project was a virtual tour of her school, built in Scratch by dragging coloured blocks into place. Then the training moved to a text editor, and the blocks became tags.
Her certificate lists what the five weeks covered: CS fundamentals, algorithms, introduction to programming, programming with Scratch, and introduction to web development — hands-on projects, mentorship and teamwork.
Understanding AI was one of three sites shipped by the web development track this summer, alongside a personal portfolio and CRISIS, a sourced explainer on the Anglophone Crisis. All three are open source at github.com/togevacm.
On the last day she stood up in front of the room, put her site on the projector, and walked everyone through it.
She had spent five weeks learning how to explain artificial intelligence to people who had never once thought about it. Then she did it out loud, to a room, at thirteen.
Lamp is thirteen. But she has already named what she wants.
During our orientation programme she told us: she wants to be a computer engineer. Not a vague interest in computers — a specific ambition, said out loud, by someone who had just spent five weeks proving she can build things.
That is exactly the kind of thing our orientation programme exists to hear, and then act on. It rests on a simple belief: a young person who can name what they want has already done the hardest part, and what they need next is someone to help them find the path. We intend to keep walking it with her.
What she has now that she did not have in July is proof — a thing she made, on the internet, with her name on it, that other people can use.
That is what the five weeks were for. Not to turn children into developers, but to move them from the side of the screen where technology happens to you, to the side where you make it.
Her own site puts it better than we can. She writes that the future will not belong only to the people with access to AI — it will belong to the people who know how to ask good questions, think critically, create responsibly, and use technology to solve problems that matter.
