Conversation starters about AI for ages 8 to 18
The best questions to ask kids about AI are open ones that ask for a guess, a test or a reason, not a fact. Ask what a machine might notice, how you could check its answer and who decided what it should do. Below are 40 questions by age, for the dinner table, the car and the classroom.
Last reviewed: 20 October 2026 7 min read
Why talk with kids about AI at all?
Because children meet AI long before anyone explains it, and a conversation is the easiest place to start. You do not need to be an expert. You need a good question and time to listen.
Many families have not had that talk yet. In a 2026 survey of more than a thousand 9- to 17-year-olds, Common Sense Media found that "nearly half of kids have never had a conversation with their parents about AI safety."
What makes a good question about AI?
A good question about AI has no single right answer, and it asks your child to think like a tester. It asks for one of three things:
| Asks for | Sounds like | Why it works |
|---|---|---|
| A guess | "What do you think it will do if…?" | A guess can be checked, so it leads straight to a test. |
| A test | "How could we find out if it's right?" | It turns an opinion into an experiment. |
| A reason | "Why do you think it did that?" | It moves from what the machine did to how it decides. |
Questions that only ask for a fact ("What does AI stand for?") end the talk quickly. Questions that ask for a guess, a test or a reason keep it going.
How do I ask so my child keeps talking?
Ask one question, then listen longer than feels natural. These habits help at every age:
- Guess first, together. Say your own guess too. It shows that guessing wrong is part of finding out.
- Say "Let's test it." When you disagree, look for a way to check, not a winner.
- Use a real machine nearby. A phone's photo search, a voice assistant, a video app's next suggestion or a spam folder all work.
- Keep "I don't know" welcome. It is an honest answer, and a good start for "How could we find out?"
- Stop while it's interesting. One good question is enough for one meal.
Questions for ages 8 to 10
At 8 to 10, keep questions concrete: one machine, one guess, one check.
At the dinner table
- 1. If a machine sorted the food on this table into "sweet" and "not sweet", what clue would it look at?
- 2. How could a computer tell a cat from a dog in a photo? What would it look for?
- 3. What would a robot get wrong about our family if it only watched us at breakfast?
- 4. If an app suggests a video you didn't like, how could it get better at choosing?
In the car
- 5. That sign tells cars what to do. If a machine read signs, what could fool it?
- 6. If a talking speaker gives a wrong answer, how would we know?
- 7. Is a calculator smart? Is a traffic light? What's the difference?
In the classroom
- 8. Sort these pictures into two groups and write the rule. Can a partner guess your rule from the groups?
- 9. A machine learned from photos of red apples only. What will it say about a green apple? Why?
- 10. Who should decide what a homework helper is allowed to say to children?
Finish with the six questions, asked about one machine you just talked about. See the six questions below.
Questions for ages 11 to 13
At 11 to 13, add counting and checking: how often is it right, and how would you know?
At the dinner table
- 11. If a chatbot sounds sure, does that mean it's right? How could you check?
- 12. When a video app chooses your next video, what do you think it is trying to get you to do?
- 13. What would you never type into an app? Why that?
- 14. If you asked a text generator to write your homework, whose work would it be?
In the car
- 15. How could a map app know there's traffic ahead? What would it need to know about other cars?
- 16. If a music app only learned from the songs you skip, what would it get wrong about you?
- 17. Would you trust a machine to pick your team for a sport? What would it need to know?
In the classroom
- 18. A model got 9 of 10 test examples right. What do you want to know about the one it got wrong?
- 19. Why should you test a machine on examples it never saw while it was learning?
- 20. A face filter works better on some faces than others. How could that happen, and how would you test it?
Finish with the six questions, asked about one app you use. See the six questions below.
Questions for ages 14 to 15
At 14 to 15, ask about trade-offs: what a system is measured on, who it fails and who answers for it.
At the dinner table
- 21. If a voice message sounded exactly like me and asked for money, how would you check it was really me?
- 22. When is using AI for schoolwork help, and when is it doing the work for you? Where's your line?
- 23. A photo looks real. What would you check before sharing it?
- 24. Should an app be allowed to learn from your messages? What would you want to know first?
In the car
- 25. A self-driving car has to choose how careful to be. Who should set that line, and how?
- 26. An AI is right 95% of the time. Is that good enough to grade tests? To check medical scans? Why the difference?
- 27. What's one thing AI does well that you'd never want it to decide alone?
In the classroom
- 28. A hiring tool learned from a company's past hires. What patterns might it copy, and how could you test it?
- 29. How could hidden instructions on a web page trick an AI tool that reads it?
- 30. If a model fails one group of people more than others, whose job is it to notice, and whose job is it to fix?
Finish with the six questions, asked about one system from the news. See the six questions below.
Questions for ages 16 to 18
At 16 to 18, ask for a defended decision: should this system be used, and on what evidence?
At the dinner table
- 31. If your school used AI to flag cheating, what evidence would you want before you trusted it?
- 32. Who should be able to see the data an app collects about you? Who decides today?
- 33. Is it fair to use a person's public posts to train an AI without asking them? What would change your mind?
- 34. What would you want to know about an AI before using it to help choose a college or a job?
In the car
- 35. A company says its AI is "safe". What would you ask them to prove, and how?
- 36. If an AI system makes a costly mistake, who should be responsible: the builder, the user or the company that sold it?
- 37. What is one job AI should never do without a person checking? Defend your answer.
In the classroom
- 38. Write a test plan for a model that sorts emails: what cases, what counts as success, and what failure costs most?
- 39. Two systems score the same overall. How would you decide which one to deploy?
- 40. Draft a one-paragraph rule for AI use in your school. Who has to agree, and how would you check it works?
Finish with the six questions, asked about a system you would or wouldn't use. See the six questions below.
What are the six questions to ask about any AI system?
Close every set with the same six questions about one machine. They are the backbone of the curriculum's learning path from 8 to 18, and every age revisits the same six questions with more depth.
- What information does the system receive?
- What does it do with that representation?
- How did its behavior arise or change?
- How do we test it on fresh cases?
- Where can it fail or be manipulated?
- Who chose the goal, who is affected, and who has authority to act?
Worked example, one machine, one family. A family asks the six questions about a video app's "up next" choice. An 8-year-old answers question 1: "It knows what I watched." An 11-year-old adds an answer to question 4: "We could watch something new and see if the next suggestion changes." A 15-year-old takes question 6: "The company chose the goal. I think the goal is keeping me watching." Nobody needed the app's code to have a real conversation about how it decides.
What this page does and does not do
It does:
- Give 40 open questions sorted by age and by setting.
- Show what makes a question about AI work: a guess, a test or a reason.
- Give six questions that work on any AI system.
It does not:
- Test, rate or compare any app or product.
- Replace your child's school rules on AI use. Check those first.
- Give legal, medical or mental-health advice. If a conversation raises a worry about your child's safety, talk to their school or a professional.
Printable summary: questions to ask kids about AI
Print this and keep it on the fridge or in the car.
Ask for a guess: "What do you think it will do if…?"
Ask for a test: "How could we find out if it's right?"
Ask for a reason: "Why do you think it did that?"
The six questions, for any machine:
- What information does the system receive?
- What does it do with that representation?
- How did its behavior arise or change?
- How do we test it on fresh cases?
- Where can it fail or be manipulated?
- Who chose the goal, who is affected, and who has authority to act?
Our question this week: ______________________________
Where does Mechna Academy fit?
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10. In each mission, a child guesses how a machine decides, then opens its Peek Hatch to check, changes one thing and tests it on a case it has never seen. Children learn how AI works this way, and they never chat with an AI.
Questions
Pick one machine you both know, such as a video app's next suggestion, and ask one open question: "How do you think it chose that?" Share your own guess, then look for a way to test it together. Listen more than you explain.
Ask questions that need a guess, a test or a reason, such as "How could we check if it's right?" or "Who decided what it should do?" Match the question to age: one machine and one check for younger children, trade-offs and evidence for teens.
Whenever your child starts noticing machines that seem to choose, such as a voice assistant or a video suggestion. Simple questions work at 8, and the same questions go deeper as children grow.
Ask for their view before giving yours, and treat them as the expert on the apps they use. Questions about trade-offs work well with teens: when is AI help fair, who is responsible for a mistake, and what evidence would make them trust a system.
What information does the system receive? What does it do with that representation? How did its behavior arise or change? How do we test it on fresh cases? Where can it fail or be manipulated? Who chose the goal, who is affected, and who has authority to act?
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10.
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10.
Start for freeThe first missions are free; membership comes after. Places are limited: we email you when your family's place opens.