AI Without the Magic

AI is not magic: it is applied math, guided by human choices. Once you see that, two things happen at once — the fear shrinks and the responsibility grows.

What this guide is for. To give you autonomy. My focus, in the classroom and here, is for each person to be able to solve their own problems with technology without depending on someone who “knows computers.” That takes two things: generative AI literacy — knowing what the machine does, where it fails, and how to ask — and computational thinking — knowing how to turn a problem into a flow, a step-by-step, layers. Generative AI democratized product building, and knowing how to use it is no longer optional, as once happened with Word, Excel, and PowerPoint. Really making the most of that still requires knowing how to think.

How it’s organized. There are three parts. In Understand, what the machine does on the inside, where bias comes from, what counts as sensitive data, and what all of this costs. In Think, how to break a problem down, give context, retrieve the right information, and decide between asking for an answer and asking for a script. In Build, how to get from a request to a product: the repository as memory, app builders, voice and images, agents on a short leash, real cases, and the responsibility of whoever ships.

Who Lia is. An invented character with a real problem. Lia works at a civil society organization and coordinates an outreach project: going after teenagers who have stopped going to school. In Brazil, millions of young people have not finished high school, and the reason most often given by those who left is needing to work (IBGE, 2026). On the other side, there are people going after them, and it works: outreach has already helped hundreds of thousands of children and adolescents get back to the classroom (UNICEF, 2025). Lia isn’t a tech person. She wants to change that reality, and she heard that “AI will solve it.”

The guide doesn’t answer for her. In each chapter there is a scene from Lia’s journey, which ends in a question; a provocation, tied to the chapter’s experiment; and a logbook, with the questions she took with her, a way to find the answer, and the piece that went into the project. There is no answer key. Lia’s answers depend on the school system she works in — and yours, on your reality. What the guide offers is the way to ask and the way to check.

How to use it. Almost every chapter has an experiment for you to play with (look for the 🧪), ends with a Take it to class box — an activity and discussion questions, to use freely — and closes with the references. If you prefer paper, there’s the book version, ready to print.

This material grew out of recordings of my classes at FGV EPPG, a school of public policy in Brasília, of conversations where I walk people through my process, and of what I’ve been writing since 2025. The text was generated with AI from that material and edited by me. It was written in Portuguese first; this is the English edition. It is a living guide: I come back, fix and expand it.