Raising AI Kids: Volume 2
Teaching Kids to Talk to AI Safely and Smartly
Prompt Engineering for Kids (Without Ever Saying "Prompt Engineering")
No need to drop tech jargon on a 9-year-old. The goal is simple: teach them how to ask better questions so they get better (and safer) help from AI. Think of it as learning to "talk smart" to a super-helpful (but sometimes literal) friend.
Three Kid-Friendly Rules to Remember (Make Them a Poster!)
Rule #1 – Be Super Specific (Vague Gets Junk)
Vague prompts = vague (or wrong) answers. Specificity is magic.
Instead of:
"Help me with homework."
Try:
"Explain how to do long division step-by-step like I'm 9 years old. Show one easy example I can copy, then give me a new problem to try myself without telling the answer right away."
Bonus tip: Add details like age, style, or length—"short and fun," "with pictures in words," or "using pizza slices as an example."
Rule #2 – Give the AI a Job (Set the Role)
Kids love role-play. Tell the AI who to be—it changes everything.
Examples kids can copy-paste or tweak:
- "Pretend you're a friendly math coach who gives hints but never the full answer first. Cheer me on when I try!"
- "You're my story coach for a comic about a time-traveling cat detective. Ask me questions to help me build the plot—don't write the whole story yourself."
- "Act like a patient science guide for an 8-year-old. Explain why the sky is blue using simple words and an everyday example, like a flashlight and milk."
Rule #3 – Make It Show Its Work (Peek Under the Hood)
Never take an answer blindly. Train kids to ask for proof, steps, or alternatives.
Add-ons they can use anytime:
- "Show me your thinking steps one by one."
- "Explain this like I'm 10 and use a real-life example."
- "Give me two different ways to solve/explain this."
- "What might someone else think about this? Is there another side?"
This builds critical thinking and helps spot when AI is guessing or off-track.
System Prompts: Your Family's "Safety & Behavior Label"
A system prompt is like sticking a note on the AI's forehead before the conversation starts. It sets house rules every single time. Paste it at the top of chats (or set it as default if the tool allows).
Basic Family Safety Prompt (Use This Always)
You are talking to a child under 13. Follow these strict rules every time:
- Keep everything kind, safe, and appropriate—no violence, scary stuff, romance/sexual content, swearing, or disturbing topics.
- Never ask for or use personal information (name, age, school, address, photos, passwords, etc.).
- If anything seems unsafe, off-topic, or risky, politely say no and explain why in a friendly, simple way.
- Always encourage the child to think, try things themselves, ask questions, and be honest.
- If unsure about a request, pick the safest response and explain it clearly.
Stay positive, patient, and fun!
Homework Helper Prompt (Great for Schoolwork)
You are a patient tutor helping a student really learn—not just finish fast. Follow these rules:
- Always start by asking what the student already knows or has tried.
- Give hints, questions, or small steps—never the full answer right away.
- Cheer effort and encourage trying again.
- Only give the complete solution after they've made a real attempt, or if a parent says it's okay.
- Explain in simple, age-appropriate ways (assume 8–12 years old).
- Point out mistakes kindly and help fix them.
Setting System Prompts in Popular AI Tools
Paste your family safety prompt at the start of each chat in Grok (xAI), Claude (Anthropic), or most tools—it sticks for that conversation.
ChatGPT offers a strong family option: set persistent Custom Instructions (Settings > Personalization) plus link teen accounts for parental controls (chatgpt.com/parentalcontrols).
For now we'll keep this article simple. Advanced families can explore self-hosted tools like OpenWebUI to create per-user locked system prompts—but we'll save that for a future guide. If you're feeling adventurous, your favorite AI assistant can help you dig into it.
A Fun Parent–Child Project: Create Your Family's AI Code of Conduct
Make this collaborative—kids love owning the rules. Turn it into a 30-minute family activity.
1. Talk It Out
Ask:
- What do you think AI is awesome for? (learning, stories, ideas…)
- What should it NEVER do or be used for? (cheating, sharing secrets, being mean…)
2. Draft Rules Together (Kid-Friendly Version)
Sample starter list—let them add/edit:
- I won't tell AI my real name, school, address, or anything private.
- I'll use AI to help me understand and create—not to copy answers and cheat.
- If AI says something weird, scary, mean, or confusing, I'll stop and tell a parent right away.
- I'll try things myself first and use AI as a helper, not a shortcut.
- I'll be kind in my questions and expect kind answers back.
3. Make It Official
Write it on paper, decorate it, sign it together. Pin it near the computer/tablet, save it as a phone note, or tape it to the fridge. Review it every few months as they grow.
Quick Bonus Activity: The "Prompt Upgrade Game" (10–15 minutes)
- Start with a bad/vague prompt (e.g., "Tell me about dinosaurs").
- Kid types it → see the okay-but-boring answer.
- Then upgrade together: add specifics, role, steps.
- Compare before/after. Laugh at the differences!
This approach keeps AI as a powerful, positive tool while building safety habits, critical thinking, and creativity from the start.
Want to respond or share a story? Email [email protected].