Artificial intelligence literacy education in primary schools: a review

Academic systematic review (primary education focus)

International (multi-country empirical literature)

Education researchers synthesizing AI literacy evidence for primary schools

• In-school (K-12)
• Informal learning

Systematic review using PRISMA-guided screening and coding

Studies from 2019 to March 2024; review published 2025

25 empirical studies (from 44 initially retrieved records)

Not single-site; tools varied across studies (intelligent agents, software, unplugged activities)

• Limited number of primary-focused empirical AI literacy studies.
• Heterogeneity in definitions, curricula, and assessment methods.
• Frequent short-term interventions and context-specific implementations limit broad generalization.

Primary school learners (young students in early and upper primary)

Mixed; many learners encounter consumer AI tools in daily life

Often limited or none, depending on study design

• Clarify how AI literacy is conceptualized in primary school contexts.
• Identify effective pedagogical and assessment approaches for young learners.
• Synthesize outcomes and challenges to guide future curriculum and policy design.

• Map theoretical frameworks underpinning existing interventions.
• Compare tool types and their roles in engagement and learning.
• Surface equity, readiness, and teacher-capacity considerations.

• Not a single intervention trial in one school.
• Not a meta-analysis with pooled effect sizes.
• Not a definitive global standard curriculum specification.

Synthesis of AI literacy learning tools and pedagogical configurations

English-language studies from Scopus and Web of Science

• Evaluator

• Database retrieval with predefined AI-literacy search terms.
• PRISMA screening with inclusion/exclusion criteria for primary contexts.
• Coding across definition, theory, pedagogy, tools, assessment, outcomes, and challenges.
• Cross-study thematic synthesis into implementation guidance.

• Transparent eligibility criteria and coding scheme.
• Inter-rater discussion process to improve coding agreement.
• Explicit reporting of limitations and future research needs.

• Identify relevant literature and apply screening criteria.
• Code studies by seven RQs covering definition through recommendations.
• Summarize frequencies of frameworks, pedagogies, tools, and assessments.
• Interpret trends, implementation barriers, and policy-level implications.

• Human researchers conducted screening, coding, and interpretation decisions.
• AI appears as learning content/tool in source studies rather than autonomous reviewer.

• Use of PRISMA structure for rigorous selection workflow.
• Evidence triangulation via mixed-method findings where available.
• Alignment of outcomes with conceptual and theoretical framing.

• Insufficient systematic primary-level AI curricula and validated measurements.
• Tool/interface limitations and computational constraints in school settings.
• Difficulty teaching abstract concepts such as data bias age-appropriately.
• Teacher professional development needs and variable student readiness.

• Frequent use of constructivist and constructionist approaches.
• Project-based, programming, and human-agent interaction strategies were emphasized.
• Use of intelligent agents and low-barrier tools to support conceptual entry points.
• Growing mixed-method evaluation to capture both performance and perception outcomes.

• Studies generally report positive motivation, engagement, and satisfaction in AI learning activities.
• Game-based and constructivist designs often increased participation and persistence.

• Academic gains include understanding core AI/ML concepts and basic data-bias awareness.
• Affective and behavioral gains include improved self-efficacy and willingness to continue AI learning.
• Soft-skill outcomes include problem solving, computational thinking, and creative expression.

Primary AI literacy should balance technical skill-building with ethics, data literacy, and inclusive pedagogies for diverse learners.

• Review emphasizes AI ethics as a core competency, not a secondary add-on.
• Critical data literacy and bias awareness are necessary for responsible AI use by young learners.
• Primary curricula should address social impact, fairness, privacy, and AI-for-social-good framing.

• Practitioner observation
• Activity documentation
• Post assessment

• Provides a primary-specific evidence map for AI literacy curriculum and assessment design.
• Identifies research gaps in longitudinal evidence, inclusivity, and validated measurement.
• Supports development of interdisciplinary and age-appropriate AI competency frameworks.

• Primary AI literacy curriculum design
• AI pedagogy and assessment in K-12
• Data literacy and AI ethics education
• Teacher readiness and professional development

• Completed

Primary school learners

Formal primary school context (plus some mixed/informal studies)

AI concept learning, human-AI interaction, and data/ethics understanding

Constructivist/constructionist, project-based, programming, and interactive methods

Medium

Medium (student-learning context with data literacy and bias-related tasks)