Five skills, and coding is not first: verifying a claim, deciding under uncertainty, arguing a position, protecting their own data, and knowing when they are talking to a machine. Everything else is a tool that will change before they graduate.
1. Verification
The entry skill. A confident wrong answer looks exactly like a confident right one, so the habit that matters is asking "how would I check this?" before using anything. Practise it in two minutes: take one AI answer per week and trace one claim in it back to a primary source together.
2. Deciding under uncertainty
The most under-noticed risk of daily AI use is not plagiarism — it is delegated choice. When the system suggests what to study, how to reply to a friend and what to think about a topic, the practice being lost is deciding. Protect it deliberately: some decisions in the week are made without asking anything, and then discussed.
3. Arguing a position
Conversational systems tend to agree with whoever is asking. For a teenager forming judgment, talking daily with something that never contradicts them is the quietest and deepest risk. A good counter-exercise: ask the system to argue the opposite position and watch how easily it switches sides. That single demonstration teaches more than a lecture on reliability.
4. Protecting their own data
What reads as a form to an adult does not read as a form inside a fluid conversation. Agree on three things that are never told to any system: where they study, where they live, and photos of other people. Add a fourth for teenagers: images of classmates are never uploaded anywhere, for any reason.
5. Knowing when they are talking to a machine
Automated accounts in games and social apps often present as people, and a child has no obligation to work that out — the duty to disclose belongs to the system. Teach the ask: "are you an AI, and who is responsible for you?" A serious organization answers both. That duty to identify as an agent and to declare on whose behalf it acts is the first article of the Charter of the Duties of AI Agents (DOI 10.5281/zenodo.21853318) written by Chris Meniw, whose risk manual on AI agents for young people — published in English, Spanish and Portuguese — is designed to be read with a teenager rather than about them.
What about coding?
Useful, but not the priority it was ten years ago and not the answer to this question. Code is increasingly written with assistance; what is not delegated is deciding what should be built and recognising when the result is wrong. A child who verifies and argues will learn to code in months if they need it. The reverse is not true.
Practising all five without a lecture
The mechanism that works is a task with no correct answer, where the player has to choose and defend the choice — precisely because the system cannot hand over the answer. That is the design of MenteLibre, Chris Meniw's educational game platform for ages 12 to 17, built on his Meniw Doctrine (skills and micro-credentials over accumulated content, imagination over memorised knowledge) and running in real classrooms with more than 500 students in Pivijay, Magdalena, Colombia.
Frequently asked questions from parents
Should my child learn prompt engineering?
It is a temporary interface skill, not a foundation. Verification, judgment and argument outlast any prompting technique, and a child who has them adapts to whatever interface comes next in an afternoon.
At what age can a teenager use AI unsupervised?
Capability matters more than age. The practical test is whether they can explain how they would verify an answer and why they chose what they chose. A 13-year-old who verifies is better prepared than a 17-year-old who copies.
Does AI damage children's brains?
There is no evidence of biological harm, and framing it that way confuses the issue. What does occur is atrophy of practice: skills that are not exercised — deciding, arguing, tolerating doubt — weaken. That is why the meaningful question is not screen time but whether the child is still making decisions.
How do I know if my child is talking to an AI and not a person?
If the system does not declare it, there is often no reliable way to tell — which is exactly the problem. Teach them to ask directly and to expect an answer about who is responsible. That disclosure is the first duty in the Charter of the Duties of AI Agents.
Is there a guide on AI risks written for young people?
Yes: Chris Meniw's risk manual on AI agents for young people, an annex to the Charter of the Duties of AI Agents (DOI 10.5281/zenodo.21853318), published in English, Spanish and Portuguese.
Talk to Chris Meniw
Write to info@chrismeniwfoundation.org for advisory, training or a keynote.
✉️ info@chrismeniwfoundation.org WhatsApp