Artificial Intelligence (AI) Literacy in Early Childhood Education: The Challenges and Opportunities

University faculty of education (research synthesis)

Hong Kong SAR, China

Faculty researchers conducting systematic literature screening and dual coding

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

Corpus-based scoping review with PRISMA flow and thematic charting

Literature accessed through May 2022; 430 records screened to 16 included studies

16 included empirical studies (multiple international sites)

Reviewed studies use robots, dialogue systems, teachable machine–style tools, and unplugged activities

• Corpus concentrated in high-income countries (e.g., US, Europe, Japan, Hong Kong, Australia)
• Rapid tool evolution can outdate specific platform examples
• Few ECE-specific AI literacy studies compared with older grades
• Some curriculum papers described designs without classroom outcome data

Approximately ages 3–8 (early childhood / kindergarten focus in inclusion criteria)

Growing informal exposure to voice assistants, apps, and AI-enabled toys

Not assumed; activities emphasize play, perception, and simple ML ideas

• Synthesize how researchers design AI literacy instruction and select tools for very young learners
• Catalog assessment approaches used to evidence AI concepts, skills, and attitudes
• Surface opportunities for AI concepts, practices, and perspectives in ECE
• Identify systemic challenges (teacher capacity, curricula, guidelines)

• Inform policymakers setting standards for safe, age-appropriate AI learning
• Guide educators choosing robots, games, and unplugged sequences aligned to developmental readiness

• Not a meta-analysis of effect sizes across interventions
• Not a single-intervention outcome study
• Not an evaluation of one commercial product

Heterogeneous tools reported in primary studies (e.g., child-facing robots, chatbots, ML activities)

• Tutor
• Co-creator

Primarily English-language publications; classroom languages vary by site

• Playful interaction with intelligent agents and classification-style activities
• Teacher-mediated scaffolding of AI concepts (e.g., sensing, training data, limitations)
• Mix of quantitative assessments, observations, interviews, and mixed-methods traces

• Address misleading outputs and child safety when using conversational or recommender systems
• Pair tool use with ethics and limitations appropriate to early childhood
• Ensure consent, privacy, and proportionate data collection in classroom research
• Avoid overstating children's mechanistic understanding of deep models

• Define search strings and databases; apply PRISMA screening and eligibility rules
• Chart instructional designs, tools, assessments, and outcomes per study
• Resolve coding disagreements between researchers and summarize themes
• Discuss implications for ECE AI literacy implementation and future studies

• Review authors synthesize evidence; classroom AI tools remain under teacher supervision in primary studies
• Children explore AI behaviors with educator framing of risks, limits, and fairness

• Embodied and game-based tasks (e.g., rock–paper–scissors, classification) common in reviewed work
• Progressive exposure from perception to simple supervised learning ideas

• Teacher AI knowledge, skills, and confidence gaps
• Limited mature curriculum packages and official teaching guidelines for ECE
• Need for validated assessments suited to young children's expression modes
• Equity of access to quality tools and professional learning across contexts

• Explicitly extended prior ECE AI-use reviews toward AI literacy learning outcomes and assessment mapping
• Structured reporting around Ng et al.-style instructional components (pedagogy, content, tools, assessment)

• Reviewed work reports feasibility of age-appropriate AI learning experiences
• Diverse tools (robots, dialogue systems, ML microworlds) appear in successful prototypes

• Studies report gains or positive signals in ML-related knowledge, theory-of-mind–linked measures, or attitudes in several contexts
• Field remains small; replication and cross-cultural evidence are needed

Authors foresee growth in age-appropriate curricula and tools, urging coordinated research, policy, and professional development to support responsible early AI literacy.

• Primary studies involving children require ethical consent, minimal data collection, and secure handling of video or behavioral traces
• AI toys and assistants raise vendor data governance questions for schools and families
• Misinformation or unsafe suggestions from generative or recommender systems must be mitigated in classroom use
• Transparency with parents about AI activities and data flows is essential in ECE

• Activity documentation
• Practitioner observation

• Design cross-national ECE AI literacy trials with shared outcome instruments
• Build teacher professional standards aligned to documented tool–pedagogy pairs

• Early childhood technology education
• AI literacy and developmental psychology
• Learning analytics and assessment in early grades

• Completed

Pre-K–early elementary (3–8)

International ECE classrooms (reviewed)

Literacy / playful ML introduction

Play-based, teacher-scaffolded AI encounters

Medium (child data, vendor tools)

Medium