AI, explained for parents
You do not need to understand the maths. You need a working picture of what these tools do, so you can use them yourself and talk about them without guessing.
Read the transcript
AI is already in your child's day. So what is it, really? A chatbot predicts the next likely word. That's why it sounds sure, even when it's wrong. Good at explaining, drafting, and practice. Bad at facts, anything recent, and knowing your child. Kids meet it in search summaries, everyday apps, homework tools, and companion apps that talk back. Only type what you'd post publicly. safernotscared.org/ai/basics.
What people mean by AI now
The AI in the news is generative AI: tools that produce new text, pictures, voices, or video when you ask. Chatbots such as ChatGPT, Gemini, Copilot, and Claude are the best known.
Older kinds of AI have been part of family life for years: the feed that picks the next video, the filter that changes a face, the voice assistant on the counter. What is new is that anyone can now ask a machine to write, draw, or talk back.
How a chatbot works, in one minute
A chatbot learned from an enormous amount of writing. When you type, it predicts which words are most likely to come next, one after another. That is why it writes so fluently, and why it can be confidently wrong: it produces what sounds right, and sounding right is not the same as being checked.
Some chatbots can also search the web and show where an answer came from. Many answers still come from the model alone, with nothing underneath them.
It is predicting, not knowing. Treat every answer like a confident first draft from a well-read stranger.
What it is good at, and what it is bad at
Good at
- Explaining a topic in simpler words, or in a different way.
- Drafting an email, a plan, a list, or a first paragraph.
- Brainstorming ideas when you are stuck.
- Summarizing something you paste in.
- Practice: quizzes, flashcards, and role-playing a hard conversation.
Bad at
- Facts it has not been given. It can invent names, numbers, quotes, and sources.
- Anything recent, unless it searches the web.
- Knowing your child, your family, or your situation.
- Judgment calls that need a person who knows what is going on.
- Being a friend. Many apps are designed to keep the conversation going, which is not the same as caring.
Where your kids already meet AI
- Search results that open with an AI-written summary.
- Chatbots built into apps they already use, including social and messaging apps.
- Homework and tutoring apps.
- Characters in games, and companion apps that talk back and remember them. These carry the most risk, explained in AI and your kid.
- Photo filters and editing tools that can change faces and bodies.
- Voice assistants and smart speakers.
Ask your child where they have seen AI this week. Most kids can name a few places their parents had not noticed.
What happens to what you type
Assume that anything typed into a chatbot is stored by the company that runs it. Many services use conversations to improve their systems unless you turn that off, and some keep a memory of you from one chat to the next. Settings differ by app and change often, so check the privacy or data settings inside the app itself.
A good family rule: do not type anything into a chatbot you would not post publicly. That includes full names, the school, the address, photos of faces, passwords, and anything about health.
Age rules differ
Some chatbots are for adults only. Some allow teenagers, often with a parent’s permission. Some offer supervised accounts for younger children through a family account. Check the terms of the specific app before your child signs up, and use your own account when you explore together with a younger child.
The fastest way to get comfortable
Use it. Try AI yourself in 20 minutes walks you through eight small exercises on your own account. Then read Use AI together, by age.
Sources
- Guidance on AI and children — UNICEF Innocenti