The Curious Generation
AI & Childhood · 2024

How Children Reason About Generative AI: A Cross-Country Study

Dr. Priya Chandrasekaran, Tomás Herrera

Abstract

What this study did

Interviewing 640 children aged 7 to 13 across four countries, this study examined how children form mental models of AI systems. Most children under 9 attributed intention and feeling to chatbots; explicit teaching about training data shifted this within six weeks.

Key findings

What we learned

01
68% of children under 9 believed a chatbot could feel proud or embarrassed.
02
A short unit on training data reduced this misattribution to 24% within six weeks.
03
Children who discussed AI at home with an adult showed more calibrated trust than those who did not.
In practice

What it means for families

Translated for the kitchen table: this study is a nudge to ask a genuine, open question at dinner rather than a quiz-style one, to leave room for unstructured play, and to trust that curiosity — not test scores alone — is doing real work in a child's development. Small, repeatable habits, not grand interventions.

Citation

How to cite this paper

Dr. Priya Chandrasekaran, Tomás Herrera (2024). How Children Reason About Generative AI: A Cross-Country Study. The Curious Generation™ Research Series.