AI literacy alone cannot prepare young children for an uncertain world

A systematic conceptual mapping review of 86 resources concludes that preparing young children for an AI-mediated world requires much more than teaching them how to use digital tools. The review, published in Frontiers in Psychology, maps the competencies, teaching practices and institutional conditions considered necessary for children aged 0 to 8, finding that AI literacy should sit within a broader framework encompassing foundational learning, emotional and social development, physical and mental well-being, executive function, and habits such as curiosity, persistence and collaboration. It also identifies play-based learning, ethical governance and sustained professional support for educators as central to age-appropriate implementation. Because the study synthesized journal papers, books, reports and policy documents rather than testing an intervention, its findings provide a conceptual framework and research agenda rather than proof that any particular classroom model produces better outcomes.

The distinction matters because modern AI systems can generate fluent, persuasive and potentially biased information. In that environment, children need to evaluate outputs rather than accept them uncritically. The review describes AI literacy as knowledge and skills for understanding, using, evaluating and creating with AI, together with attitudes and ethical understanding. Proposed frameworks variously organize this capability into knowledge, skills and attitudes, or into safety, identity, attitude, cognition and capability. These dimensions depend on capacities developed earlier: working memory, inhibitory control and cognitive flexibility help children hold information, suppress impulsive responses and adapt their attention, while critical thinking and creativity support the identification of bias and the generation of alternatives. Early AI education has been associated in the reviewed empirical studies with computational thinking, sequencing, self-regulation and theory of mind, although the strength and transferability of those associations remain questions for direct testing.

Foundational literacy and numeracy remain central, but the review argues that early childhood development must also nurture resilience, adaptability, motivation, self-awareness, empathy and the ability to understand other people’s perspectives. Theory of Mind, described as the capacity to attribute beliefs, desires, intentions and knowledge to oneself and others, receives particular attention because human understanding is grounded in emotion and lived social experience, unlike the pattern-based operation of large language models. Learning dispositions—characteristic ways of approaching situations—help determine whether children persist, explore and cooperate when an unfamiliar technological activity is confusing or uncertain. The review treats these abilities as mutually influential rather than separate targets: well-being supports sustained attention, executive function supports self-directed learning, and curiosity and persistence can promote critical and ethical engagement with technology.

The synthesis also warns that digital exposure can affect children beyond the classroom. Reviewed literature associates excessive screen time with increased risks of myopia, while prolonged screen exposure is linked with eye strain, sleep quality, physical activity and mental health, alongside inappropriate-content and online-safety risks. These are reported associations in the source material, not estimates generated by the present review, and they do not establish that every AI-supported educational activity causes harm. The proposed response is ecological: digital service providers, health professionals, educators, parents and children would all participate in creating safer digital conditions. Embedding physical tools with digital functions is one possibility discussed in the literature, but the review does not establish a single intervention as effective or superior.

Governance must address children’s developmental characteristics alongside technical performance. Young children may treat robots and AI characters as social agents, creating concerns about over-trust and blurred boundaries between people and machines. The reviewed policy documents emphasize privacy, transparency, equity, inclusion, well-being and responsible use, with Australian, Chinese, United States, OECD and UNESCO materials illustrating how governance is shifting from unrestricted innovation toward child-centred safeguards. Recommendations include child-specific datasets, algorithmic transparency, accurate communication about system capabilities and ways for children’s voices to influence design. The review also notes that blanket prohibitions may not eliminate generative AI; they may instead push use outside visible school oversight. Governance therefore requires both permission to experiment and enforceable limits on privacy, bias, inequitable access and inappropriate human–machine relationships.

Early childhood educators face particular pressures as AI changes their professional role. According to material included in the review, women constitute 94% of pre-primary teachers worldwide, bringing both the workforce’s digital literacy gaps and workplace stress into focus. Technical training alone may not prepare teachers to evaluate generated materials, interpret errors or make ethically appropriate decisions about young children. The literature instead emphasizes critical engagement, digital confidence, self-efficacy, mentoring and practical experience that connect AI capabilities to developmentally appropriate pedagogy. Educators are portrayed not as obsolete content providers but as facilitators, co-designers and ethical overseers who contextualize algorithmic outputs. Adaptive systems may assist with personalization or workload, but they still require teacher oversight, and excessive automation risks narrowing professional agency. Research on educators’ AI literacy remains limited in the corpus reviewed.

Participatory design offers a mechanism for improving both implementation and teacher agency. When educators help design AI-supported activities, they can test whether tools fit local values, classroom routines and children’s developmental needs while feeding practical experience back into institutional policy. Design-based research included in the review indicates that co-design can strengthen teacher agency and support ethically and developmentally appropriate interventions. The same principle applies to professional learning generally: hands-on workshops and mentorship may help educators incorporate tools into teaching, while collaboration among schools and families can provide children with consistent guidance on ethical use. The review proposes these as directions supported by the mapped literature, not as interventions whose effectiveness has been established through a controlled trial within this study.

At the level of direct learning, the evidence supports a role for AI that supplements rather than replaces children’s activity, relationships and agency. High-quality early childhood education centers on open-ended play, storytelling, collaboration, exploration and care, and the review argues that technology should extend those experiences. Tangible and unplugged activities can introduce computational ideas, while block-building applications can connect artificial intelligence with creative problem-solving in sustainable city design. AI-enabled role-play and educational robots have supported learning and inquiry literacy in the reviewed studies, including programs using direct instruction with scaffolding and cooperative robot play. One eight-week curriculum indicated that direct instruction with scaffolding improved certain knowledge and skills, whereas cooperative play better supported positive attitudes. These findings come from separate programs and should not be generalized to all AI tools, classrooms or children.

The review’s main contribution is therefore an ecological synthesis rather than a tested curriculum. Drawing on Bronfenbrenner’s ecological systems theory, it links children’s direct experiences with educators and peers to connections among families and schools, institutional and community influences, wider cultural and policy environments, and changes over time. AI literacy is consequently repositioned as one component of a mutually reinforcing developmental system supported by relationships, governance and professional capacity. The authors acknowledge several limitations: most evidence came from high- and middle-income settings, AI technologies were evolving rapidly, and thematic interpretation remains partly subjective. Coding was also conducted by a single researcher without assessment of inter-coder reliability, although the framework was refined through verbatim excerpts and coding memos. Future work should move beyond conceptual mapping to empirical and design-based studies across diverse settings to determine how AI integration is actually enacted, for whom it works, and under what institutional conditions.

Subject of Research: A conceptual mapping of AI-related competencies, pedagogy and systemic support in early childhood education for children aged 0 to 8.

Article Title: Shaping early childhood in the age of AI: holistic competencies, pedagogical practices, and ecological support

Article References: Pei, L. (2026). Shaping early childhood in the age of AI: holistic competencies, pedagogical practices, and ecological support. Frontiers in Psychology, 17, Article 1995020. https://doi.org/10.3389/fpsyg.2026.1995020

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1995020

Keywords: early childhood education, artificial intelligence, AI literacy, play-based learning, executive function, young children, educator development, ethical governance, digital well-being, ecological systems theory, Shaping, early