Evidence

A promising direction. Clear limits.

Thirty-two sources, each with what it found, what it can’t tell us, and a classroom move. Research on how young children learn from people and characters is not research on chatbots, and these activities haven’t been tested as an AI-literacy intervention.

Sources

We can sayGuided play has benefits for some outcomes. Young children weigh the reliability of speakers in some tasks. Children form meaningful bonds with media characters and learn more from characters they feel close to. Pretend friends are common; individual relationships and timelines vary. Children often overestimate what smart devices understand.

We can’t sayThat these activities cause later AI literacy, that toddlers understand algorithms, that any product is safe for young children, or that a character’s bond is risk-free.

Checked in October 2026 against publisher pages, abstracts or the researchers’ own published summaries. Each card’s “Access” line says what was reviewed; where only an abstract or summary was available, the card sticks to what it reports. Peer-reviewed research sits alongside rights guidance and two consumer reports, labelled as such.