AI Pre-Literacy Research Observatory

Early Childhood and AI: Evidence Map

A searchable evidence map for research on young children’s encounters with artificial intelligence and related intelligent agents. This is a living evidence infrastructure; the peer-reviewed synthesis reports a locked snapshot.

Methodological Note — Locked Universe 21 Aug 2026. The Observatory is a living evidence infrastructure. This view shows a locked discovery universe of 62 records as of 21 August 2026. It contains 15 verified primary empirical records (children approx. 2–8 years, child-level evidence, peer-reviewed), 9 records on Hold pending age/full-text verification, 16 contextual records (reviews, parent/design evidence, international reports), 12 excluded records, and 10 duplicate/merged records. Counts refer to evidentiary role, not quality ranking. Original study terminology is preserved; AI pre-literacy is introduced only at synthesis. The historical 21/18/1/7/15 counts are superseded by this row-level reconciliation. Publication-status correction: Kim et al. (#39/#52) is now peer-reviewed primary evidence (published June 2026).

Evidence Map

Search and filter the research landscape. Select See Analytical Review for the detailed study record. Use Status filter to view Includes, Holds, Contextual, Excluded, Duplicate.

62Discovery records (immutable)

15Verified Primary — Include

9Pending Verification — Hold

16Contextual

12Excluded

10Duplicate / Merged

3Analytical Layers

21 Aug 2026Last Locked

Study	Year	Age	Technology / agent	Developmental phenomenon	Relationship	Layer	Status

Version and Audit Trail — Locked Universe

Locked discovery pool: 62 records, immutable, search 2021 — 21 Aug 2026. No new broad-search records added after lock.
Current reconciliation: AI_Pre_Literacy_62_Record_Reconciliation_Matrix.xlsx — 15 Include / 9 Hold / 16 Contextual / 12 Exclude / 10 Duplicate/Merged — file available for download: CC BY 4.0.
Age rule: Chronological age must be verified. School grade is NOT used as age proxy. Example: Kyaw et al. (#35) Grade 2 remains Hold.
Duplicate rule: Kim #39 = #52 (DOI: 10.1080/10447318.2026.2678532) retained as #39; Kosoy #40 = #53 (arXiv:2405.13081) retained as #40; Neumann #42 = #30; #44 = #31; #57 = #45; #59 = #43; #60 = #23; #61 = #14; #62 = #19; Su & Yang #56 = merged contextual.
Publication-status correction: Kim et al. published June 2026, within locked window, now peer-reviewed primary.
Download — CC BY 4.0: Excel — Full Reconciliation Matrix (62 records, 23 fields) | CSV — Public Version (62 records, 16 fields) | DOI: to be minted on Zenodo at publication
License: CC BY 4.0 — you may reuse with attribution. Cite as: Pasha, A.U.R. (2026). AI Pre-Literacy 62-Record Reconciliation Matrix. Locked Universe 21 Aug 2026. aipreliteracy.org/research-observatory/
Archive: Wayback Machine snapshot recommended for timestamped verification — https://web.archive.org/save/https://aipreliteracy.org/research-observatory/

About the Evidence Map

The Observatory is designed as evidence infrastructure rather than a conventional bibliography. Existing research retains its original terminology and theoretical identity. The proposed AI pre-literacy domain is introduced only at the synthesis level as a provisional organizing domain. Peer-reviewed empirical research forms the principal primary evidence set; contextual, emerging, duplicate, excluded, and unresolved records are kept analytically distinct.

Three-Layer Analytical Boundary

Layer 1 — Core Developmental Territory

AI recognition and vocabulary; emerging concepts and mental models; perceptions and attitudes; agency attribution and anthropomorphism; imaginative meaning-making. Most direct processes of recognition and sense-making.

Layer 2 — Developmental Extension

Social interaction; emotional responses; trust and relational orientation; behavioural responses; autonomy; emerging socio-ethical awareness. Associated responses and orientations, not stages.

Layer 3 — Developmental Ecology and Context

Home and family; preschool; peers; community; culture and language; socioeconomic conditions; media and technology environments; adult mediation; AI-enabled toys and devices; social norms. Context is ecological condition, not component of AI pre-literacy itself.

Analytical, not developmental: These layers are analytical categories, not developmental stages, hierarchical levels, or validated components of a measurement model. Layer 1 concerns the most direct processes; Layer 2 captures extensions; Layer 3 identifies contexts.

Methodological Position

The evidence map uses a broad, multi-concept discovery search and a structured verification process covering AI and early-childhood terminology, social robotics, intelligent agents, developmental psychology, trust, anthropomorphism, and agency attribution (2021 — 21 Aug 2026, English-language initial review). Peer-reviewed empirical research forms the principal primary evidence set; contextual, emerging, duplicate, excluded, and unresolved records are kept analytically distinct. The Observatory does not treat all social robots as AI, and school grade is not used as a proxy for chronological age. Original terminology is preserved throughout; AI pre-literacy is introduced only at synthesis, avoiding circular selection.

Boundary — What Is Excluded

Teleoperated / Wizard-of-Oz robots without AI involvement, or where AI status unclear and child-AI relationship not investigated (e.g., Tolksdorf #1)
Robot-mediated play / physical activity where child-AI relationship is not substantive (e.g., Raja Vora #3)
Language-learning effectiveness / tutoring effectiveness as principal outcome (e.g., van den Berghe #2, Pareto #10)
Studies outside 2-8 years where target age is not separable (e.g., Heeg & Avraamidou #23 age 11-12; Maj #58 age 7-12)
Computational thinking, coding, programming as principal focus
Adult-only, parent-only, manufacturer-claim, or design-intention without child-level evidence (retained as Contextual, not Primary)