How to protect your child's thinking skills from AI, age by age

Protect your child's thinking by keeping the hard part with them: they try first, AI only gives hints or asks questions, and they explain the answer back without the screen. Pair that with lots of AI-free time and an adult who stays involved, and AI becomes a sparring partner rather than a substitute.

Written by the Ask Poppy team, who make one of the tools mentioned here. The advice works with any AI, or none. We haven't run a controlled study of our own app.

The core idea: who does the thinking?

Learning happens when a child retrieves, explains, struggles and corrects. Research on "self-explanation" (explaining steps to yourself) finds a reliable benefit across 69 effects (Bisra et al., 2018). When AI does those steps instead, practice looks better but learning can drop: in a PNAS 2025 high-school study, unrestricted ChatGPT help left students 17% worse on a later test without AI, while a hint-giving version didn't (Bastani et al., 2025). So the question to ask about any AI use is simple: who's doing the thinking?

Age-by-age plan

Ages 3 to 5: talk, play, no solo AI

WHO advises no sedentary screen time for children under 2 and no more than an hour a day at 2 to 4, with less being better, and encourages reading and storytelling with a caregiver (WHO, 2019). At this age the "thinking skill" is conversation itself. If AI shows up at all, it's you asking it a question together, out loud, and then going to look at the real thing.

Ages 6 to 8: AI asks, the child answers

• Short voice sessions (10 to 15 minutes) with you in the room.
• The AI's job is to ask questions: "What do you think happens next?" The child does the reading, counting and drawing.
• Rule: "We find out together." Check one AI claim a week in a book or by trying it.

Ages 9 to 12: try first, then hints

• The two-minute rule: a real attempt before asking AI.
• Hints one at a time, never the full solution first. Our hint ladder gives you the prompt.
• Teach that AI can be wrong. Spot-check answers with a two-minute check.

Ages 13 to 16: feedback on their own work

• AI gives feedback on drafts the teen wrote, quizzes them before tests, or argues the other side.
• They follow school rules on AI and say when they used it (see is using AI for homework cheating?).
• Watch for overconfidence: adults using AI overrated their own performance (Fernandes et al., 2026).

Family routines that work

• Explain-it-back at dinner. "Teach me one thing you learned today." No screen.
• Paper first. Plans, outlines and maths working start on paper.
• AI in shared spaces. Kitchen table, not bedroom.
• Boredom is allowed. Don't fill every gap with a device; boredom is where children invent.
• Model it. Let them see you try first, check AI claims and disagree with it.

Warning signs

• They reach for AI before reading the question.
• They can't explain what they handed in.
• Their writing suddenly doesn't sound like them.
• They get upset without AI on tasks they used to manage.

What to look for in an AI tool

On that last point, Common Sense Media rated "social AI companions" (apps built as AI friends or partners) an unacceptable risk and recommends no one under 18 use them (Common Sense Media, 2025). A learning tool should feel like a tutor, not a best friend.

Where Ask Poppy fits

We built Ask Poppy around these ideas. It's a voice-first companion: a child holds to talk, and Poppy asks what they already know and offers one hint at a time instead of leading with the answer. Parents can add up to three child profiles, open each child's conversations (children are told a parent can see them), and turn memory off per child. A one-button, screen-free Poppy device is in development. You can reserve one on askpoppy.ai with a fully refundable $30 toward a planned $200 price; estimated shipping is spring 2027. The app works fully without it. Like any AI, Poppy can make mistakes, which is one more reason to keep an adult involved.

Sources (checked 26 Oct 2026): WHO guidelines for children under 5, Apr 2019 · Bastani et al., PNAS, June 2025 · Common Sense Media, AI companions risk assessment, Apr 2025 · Fernandes et al., Computers in Human Behavior, 2026. Our plain-English research summaries are on research.askpoppy.ai. Spot something out of date? Email [email protected].

Questions

How can I protect my child's thinking skills from AI?

Keep the hard part of the thinking with your child: they attempt first, AI gives hints or asks questions, and they explain the result back without the screen. Add plenty of AI-free reading, play and conversation, and stay involved in how and when AI is used.

At what age should a child use AI?

There's no official age for AI specifically. WHO advises no sedentary screen time under 2 and at most an hour a day at 2 to 4. Many general chatbots, including ChatGPT, require users to be 13 or older with parental consent. Before then, keep any AI use short, talk-based and with an adult.

Which thinking skills are most at risk from AI?

The ones AI does most easily: writing and organising ideas, recalling facts, working through maths steps and judging whether information is reliable. Studies of teens and adults found that handing these over lowered learning or reasoning.

Should I ban AI completely?

You don't have to. Bans are hard to hold as children get older, and studies show AI that asks questions and gives hints can help learning. A better rule is "AI can help you think, not think for you", with you involved.

How do I know if AI is replacing my child's thinking?

Ask them to explain the homework without the screen. If they can't, can't remember what they wrote, or reach for AI before trying, it's doing too much.

More on kids and AI

• Does AI make kids dumber? What the research shows
• Cognitive debt: what AI can do to kids' thinking
• Does AI hurt kids' writing and creativity?
• A Waldorf-inspired approach to AI at home
• Screen-free AI for kids: voice first, screens later
• All guides on raising kids alongside AI
• AI tutors compared side by side

By Simon Dreyer, founder of Ask Poppy · Last updated 26 October 2026 · How we write these pages