A critical review of 34 studies reveals that early childhood artificial intelligence research prioritizes cognitive and situated perspectives while lacking critical frameworks addressing societal equity and posthumanist ethics.
Overview of the Critical Review A comprehensive critical review published in AI, Brain and Child systematically evaluated how artificial intelligence (AI) is conceptualized in early childhood education (ECE) research. The investigation utilized cognitive, situated, and critical computational thinking frameworks to analyze empirical and theoretical publications. Researchers identified a significant imbalance in the existing literature, noting a heavy concentration on individual cognitive gains and situated environmental interactions, but a conspicuous absence of critical perspectives addressing systemic equity, societal consequences, and democratic access. To address these gaps, the authors advocate for the integration of a posthumanist perspective and an ethic of care framework to cultivate socially responsible and sustainable early learning ecosystems. The review adhered to systematic search and selection principles outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Literature searches were conducted across Google Scholar, Web of Science, and Scopus, covering publications through June 2024. The retrieval strategy involved keyword combinations of AI terminology, early childhood age ranges, and foundational learning theories. Studies focusing on special education, teacher/parent training, or non-peer-reviewed commentaries were excluded to maintain a focus on direct educational impacts on typically developing young learners. The initial search yielded 579 records, which were narrowed down to 301 after title and abstract screening, and further reduced to 159 candidates through secondary evaluation. A final appraisal, based on explicit theoretical integration criteria, identified 34 qualifying papers, consisting of 21 empirical studies and 13 conceptual or review articles. Among these, 8 articles aligned with cognitive framing, 17 with situated framing, and 10 with critical framing, with one study exhibiting characteristics of both cognitive and situated domains. Cognitive Framing: Understanding AI's Impact on Children's Mental Operations The cognitive framing, represented by eight studies, focused on children's internal mental operations such as reasoning, memory, attention, and motivation. This approach drew heavily from cognitivist principles, particularly those of Jean Piaget and Susan Carey. Empirical research within this domain primarily employed quantitative designs with larger samples to evaluate children's perceptions of animacy in technological agents. These studies were often grounded in concepts of animism, initially articulated by Robert Ranulph Marett and further developed by Piaget, assessing how children categorize interactive machines. For instance, Tanya Beran and colleagues conducted a study with 184 participants aged 5 to 16, revealing that younger children frequently attributed cognitive abilities, such as memory, to robots even when they were manually controlled. Affective attributions, in contrast, were observed to persist across all age groups. Similarly, Darren Cameron and co-authors examined 72 children aged 4 to 12, finding that children aged six and under tended to perceive humanoid robots as person-like when they exhibited autonomous behavior, whereas older children more readily classified them as mechanical devices. Expanding on these cognitive categorizations, Miri Kim and colleagues evaluated animacy reasoning among 120 children aged 3 to 5 who interacted with humanoid robots under four conditions varying in mobility and contingency. Their findings indicated that three-year-olds predominantly relied on movement cues to determine if an agent was alive, while five-year-olds prioritized psychological contingency and responsiveness. Beyond animacy, cognitive frameworks also integrated self-determination theory and developmental stage models. Ting Zhang and Weiqin Tang, for example, utilized Piagetian stages to structure three distinct companion robot interaction modes—visual displays, speech recognition, and physical gestures—tailored for learners aged zero to eight. Additionally, Paulien van Minkelen and colleagues studied 49 Dutch children aged 5 to 6, observing that self-determination relatedness support enhanced task engagement during second-language learning, though it did not directly improve vocabulary acquisition. Conceptual papers by Jiahong Su, Weipeng Yang, and Yuchun Zhong translated abstract artificial intelligence concepts into early childhood curricula, incorporating playful role-play and structured competencies that spanned knowledge, skills, and ethical attitudes. Situated Framing: AI in Sociocultural Learning Environments Situated framings emerged as the most prevalent category in the review, encompassing 17 articles that explored the dynamic interplay between technological agents, learners, and their sociocultural environments. This body of literature was deeply rooted in Lev Vygotsky’s sociocultural theory, emphasizing how social and linguistic interactions mediate cognitive growth, particularly through the concept of the Zone of Proximal Development. Numerous investigations demonstrated the capacity of artificial intelligence tools to scaffold learning tasks, adapting to match children’s emerging capabilities, as evidenced by research conducted by Antonio Silva and colleagues, as well as Fredy Baskara. Observational and experimental studies further confirmed the relevance of Albert Bandura’s social learning theory in this context. Qianqian Jia and collaborators, for example, showed that robots modeling prosocial behavior significantly increased sharing tendencies among young children. Moreover, comparative experimental work by Ömer Demir-Lira and colleagues identified that visual cues presented by social robots were more effective than physical gestures in teaching language concepts. Participatory activities, such as those employing Google’s Teachable Machine, enabled active knowledge construction through collaborative exploratory engagement. Situated research also illuminated the functional roles of artificial intelligence and the mediation processes that shape collaborative learning environments. Drawing from George Herbert Mead’s social theory, Katharina Rohlfing and colleagues demonstrated that social robots, when configured as tutors, peers, or novices, fundamentally alter interactive behaviors, the depth of engagement, and learner expectations. Applying mediation theory, studies by Sarika Kewalramani and co-authors illustrated how robotic toys, when appropriately scaffolded by educators, fostered collaborative inquiry and creative exploration in early classrooms. Additionally, studies rooted in multimodality and Seymour Papert’s constructionism highlighted how embodied interactions reinforce conceptual learning. Cristina Sabena documented how programmable robots facilitate spatial reasoning through physical navigation tasks, while Ilene Berson and colleagues analyzed multimodal creative inquiry utilizing tools like Citizen DJ to combine musical and kinetic exploration, showcasing how sensory-rich environments promote meaningful early learning experiences. Critical Framing: Addressing Ethics, Equity, and Posthumanist Perspectives In stark contrast to the prevalence of cognitive and situated framings, the critical framing remained significantly underdeveloped, accounting for only 10 articles. This literature primarily addressed critical issues such as ethics, power structures, social equity, and democratic participation within the context of AI in early childhood education. AI was often conceptualized through metaphors of dual impact, such as a "double-edged sword" or "Pandora’s box," highlighting the balance between personalized learning affordances and the risks of data privacy violations, algorithmic bias, and digital dependency. To guide pedagogical practice in this complex landscape, Jiahong Chen and Jiangchun Lin introduced the POWER principles, advocating for purposeful, optimal, wise, ethical, and responsible integration of AI. Similarly, Wei Luo and colleagues formulated the 3A2S framework, addressing accessibility, affordability, accountability, sustainability, and social justice. Other studies focused on nurturing critical technological reflection among children themselves. Blandine Fanny and co-authors investigated metaphorical discourses to cultivate early robot literacy, encouraging children to think critically about AI. Nils Tolksdorf and colleagues demonstrated that interactions with social robots stimulated 'metatalk' and metacognitive reflection in preschool classrooms. However, the authors emphasized that empirical critical research currently lacks longitudinal, child-centered designs necessary for deeper insights. Critical perspectives also interrogated systemic equity and explored non-human relational ethics within early childhood spaces. Addressing educational disparities, Helena Pozniak argued that AI algorithms could potentially help identify students requiring targeted intervention to mitigate educational exclusion, though rigorous human oversight remains imperative to prevent automated systems from reinforcing institutional biases. Concurrently, Weipeng Yang and colleagues evaluated robot programming initiatives in under-resourced schools, revealing how hardware scarcity and insufficient teacher training severely impede equitable technological access. Theoretical contributions challenged traditional boundaries by framing robots as a new ontological category, a concept explored by Peter Kahn and colleagues. Research consistently showed that children often perceive social robots simultaneously as artifacts and social companions, prompting complex moral judgments. Studies by Daniel Martin and colleagues, alongside work by Kristin Brink and Henry Wellman, demonstrated that children extend prosocial norms, altruism, and and trust towards technological agents. Furthermore, Denise Silvis and collaborators illustrated that children caring for malfunctioning robots cultivated mutual empathy and reciprocal responsibility. Synthesis and Global Implications for Early Childhood Education Synthesizing these findings, the critical review effectively mapped how each framing corresponds to foundational curriculum questions in early childhood education. Research within the cognitive framing primarily addresses 'what' content children should learn, focusing on computational concepts and foundational domain knowledge. The situated framing addresses 'how' pedagogical experiences should occur, promoting culturally responsive, play-based, and embodied methods that effectively leverage contextual environments. The critical framing, though underdeveloped, directly engages the sociotechnical question of 'who benefits,' seeking to dismantle systemic oppression, bridge digital divides, and protect marginalized populations. To resolve inherent tensions across these diverse dimensions, the review authors propose adopting a posthumanist theoretical lens. This philosophical perspective critiques anthropocentric assumptions, viewing artificial intelligence not merely as an instrument for human utility, but as an agential material entity engaged in intra-action within broader social and ecological networks. This approach aligns with Rosi Braidotti’s affirmative ethics of mutual interdependence and ecological frameworks that position artificial intelligence within a context of planetary coexistence. Finally, the critical review highlighted substantial geographic imbalances in the existing literature. It noted that research is overwhelmingly centered on Western contexts, leaving noticeable empirical gaps regarding early childhood traditions and AI integration across the Global South. Analyzing research conducted in Asian settings, scholars such as Jing Xu identified that regional studies frequently remain broad without achieving sufficient empirical depth. Furthermore, pedagogical traditions, such as East Asian Confucian teaching models, which emphasize teacher-directed authority, present distinctive adaptation challenges for child-centered artificial intelligence applications. However, the authors noted that posthumanist ethics and an ethic of ecological care naturally resonate with long-standing Asian philosophies, including Buddhist and Daoist principles that prioritize universal harmony and relational interdependence among living and non-living entities. For early childhood practitioners, the review offers crucial practical implications: it recommends adopting developmentally tailored AI tools that match cognitive capacities, utilizing multimodal robots to foster collaborative social expression, maintaining ethical transparency regarding machines' psychological limits, and ensuring equitable resource distribution to prepare young children for an increasingly automated and interconnected society.
Overview of the Critical Review
A comprehensive critical review published in AI, Brain and Child systematically evaluated how artificial intelligence (AI) is conceptualized in early childhood education (ECE) research. The investigation utilized cognitive, situated, and critical computational thinking frameworks to analyze empirical and theoretical publications. Researchers identified a significant imbalance in the existing literature, noting a heavy concentration on individual cognitive gains and situated environmental interactions, but a conspicuous absence of critical perspectives addressing systemic equity, societal consequences, and democratic access. To address these gaps, the authors advocate for the integration of a posthumanist perspective and an ethic of care framework to cultivate socially responsible and sustainable early learning ecosystems. The review adhered to systematic search and selection principles outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Literature searches were conducted across Google Scholar, Web of Science, and Scopus, covering publications through June 2024. The retrieval strategy involved keyword combinations of AI terminology, early childhood age ranges, and foundational learning theories. Studies focusing on special education, teacher/parent training, or non-peer-reviewed commentaries were excluded to maintain a focus on direct educational impacts on typically developing young learners. The initial search yielded 579 records, which were narrowed down to 301 after title and abstract screening, and further reduced to 159 candidates through secondary evaluation. A final appraisal, based on explicit theoretical integration criteria, identified 34 qualifying papers, consisting of 21 empirical studies and 13 conceptual or review articles. Among these, 8 articles aligned with cognitive framing, 17 with situated framing, and 10 with critical framing, with one study exhibiting characteristics of both cognitive and situated domains.
Cognitive Framing: Understanding AI's Impact on Children's Mental Operations
The cognitive framing, represented by eight studies, focused on children's internal mental operations such as reasoning, memory, attention, and motivation. This approach drew heavily from cognitivist principles, particularly those of Jean Piaget and Susan Carey. Empirical research within this domain primarily employed quantitative designs with larger samples to evaluate children's perceptions of animacy in technological agents. These studies were often grounded in concepts of animism, initially articulated by Robert Ranulph Marett and further developed by Piaget, assessing how children categorize interactive machines. For instance, Tanya Beran and colleagues conducted a study with 184 participants aged 5 to 16, revealing that younger children frequently attributed cognitive abilities, such as memory, to robots even when they were manually controlled. Affective attributions, in contrast, were observed to persist across all age groups. Similarly, Darren Cameron and co-authors examined 72 children aged 4 to 12, finding that children aged six and under tended to perceive humanoid robots as person-like when they exhibited autonomous behavior, whereas older children more readily classified them as mechanical devices. Expanding on these cognitive categorizations, Miri Kim and colleagues evaluated animacy reasoning among 120 children aged 3 to 5 who interacted with humanoid robots under four conditions varying in mobility and contingency. Their findings indicated that three-year-olds predominantly relied on movement cues to determine if an agent was alive, while five-year-olds prioritized psychological contingency and responsiveness. Beyond animacy, cognitive frameworks also integrated self-determination theory and developmental stage models. Ting Zhang and Weiqin Tang, for example, utilized Piagetian stages to structure three distinct companion robot interaction modes—visual displays, speech recognition, and physical gestures—tailored for learners aged zero to eight. Additionally, Paulien van Minkelen and colleagues studied 49 Dutch children aged 5 to 6, observing that self-determination relatedness support enhanced task engagement during second-language learning, though it did not directly improve vocabulary acquisition. Conceptual papers by Jiahong Su, Weipeng Yang, and Yuchun Zhong translated abstract artificial intelligence concepts into early childhood curricula, incorporating playful role-play and structured competencies that spanned knowledge, skills, and ethical attitudes.
Situated Framing: AI in Sociocultural Learning Environments
Situated framings emerged as the most prevalent category in the review, encompassing 17 articles that explored the dynamic interplay between technological agents, learners, and their sociocultural environments. This body of literature was deeply rooted in Lev Vygotsky’s sociocultural theory, emphasizing how social and linguistic interactions mediate cognitive growth, particularly through the concept of the Zone of Proximal Development. Numerous investigations demonstrated the capacity of artificial intelligence tools to scaffold learning tasks, adapting to match children’s emerging capabilities, as evidenced by research conducted by Antonio Silva and colleagues, as well as Fredy Baskara. Observational and experimental studies further confirmed the relevance of Albert Bandura’s social learning theory in this context. Qianqian Jia and collaborators, for example, showed that robots modeling prosocial behavior significantly increased sharing tendencies among young children. Moreover, comparative experimental work by Ömer Demir-Lira and colleagues identified that visual cues presented by social robots were more effective than physical gestures in teaching language concepts. Participatory activities, such as those employing Google’s Teachable Machine, enabled active knowledge construction through collaborative exploratory engagement. Situated research also illuminated the functional roles of artificial intelligence and the mediation processes that shape collaborative learning environments. Drawing from George Herbert Mead’s social theory, Katharina Rohlfing and colleagues demonstrated that social robots, when configured as tutors, peers, or novices, fundamentally alter interactive behaviors, the depth of engagement, and learner expectations. Applying mediation theory, studies by Sarika Kewalramani and co-authors illustrated how robotic toys, when appropriately scaffolded by educators, fostered collaborative inquiry and creative exploration in early classrooms. Additionally, studies rooted in multimodality and Seymour Papert’s constructionism highlighted how embodied interactions reinforce conceptual learning. Cristina Sabena documented how programmable robots facilitate spatial reasoning through physical navigation tasks, while Ilene Berson and colleagues analyzed multimodal creative inquiry utilizing tools like Citizen DJ to combine musical and kinetic exploration, showcasing how sensory-rich environments promote meaningful early learning experiences.
Critical Framing: Addressing Ethics, Equity, and Posthumanist Perspectives
In stark contrast to the prevalence of cognitive and situated framings, the critical framing remained significantly underdeveloped, accounting for only 10 articles. This literature primarily addressed critical issues such as ethics, power structures, social equity, and democratic participation within the context of AI in early childhood education. AI was often conceptualized through metaphors of dual impact, such as a "double-edged sword" or "Pandora’s box," highlighting the balance between personalized learning affordances and the risks of data privacy violations, algorithmic bias, and digital dependency. To guide pedagogical practice in this complex landscape, Jiahong Chen and Jiangchun Lin introduced the POWER principles, advocating for purposeful, optimal, wise, ethical, and responsible integration of AI. Similarly, Wei Luo and colleagues formulated the 3A2S framework, addressing accessibility, affordability, accountability, sustainability, and social justice. Other studies focused on nurturing critical technological reflection among children themselves. Blandine Fanny and co-authors investigated metaphorical discourses to cultivate early robot literacy, encouraging children to think critically about AI. Nils Tolksdorf and colleagues demonstrated that interactions with social robots stimulated 'metatalk' and metacognitive reflection in preschool classrooms. However, the authors emphasized that empirical critical research currently lacks longitudinal, child-centered designs necessary for deeper insights. Critical perspectives also interrogated systemic equity and explored non-human relational ethics within early childhood spaces. Addressing educational disparities, Helena Pozniak argued that AI algorithms could potentially help identify students requiring targeted intervention to mitigate educational exclusion, though rigorous human oversight remains imperative to prevent automated systems from reinforcing institutional biases. Concurrently, Weipeng Yang and colleagues evaluated robot programming initiatives in under-resourced schools, revealing how hardware scarcity and insufficient teacher training severely impede equitable technological access. Theoretical contributions challenged traditional boundaries by framing robots as a new ontological category, a concept explored by Peter Kahn and colleagues. Research consistently showed that children often perceive social robots simultaneously as artifacts and social companions, prompting complex moral judgments. Studies by Daniel Martin and colleagues, alongside work by Kristin Brink and Henry Wellman, demonstrated that children extend prosocial norms, altruism, and and trust towards technological agents. Furthermore, Denise Silvis and collaborators illustrated that children caring for malfunctioning robots cultivated mutual empathy and reciprocal responsibility.
Synthesis and Global Implications for Early Childhood Education
Synthesizing these findings, the critical review effectively mapped how each framing corresponds to foundational curriculum questions in early childhood education. Research within the cognitive framing primarily addresses 'what' content children should learn, focusing on computational concepts and foundational domain knowledge. The situated framing addresses 'how' pedagogical experiences should occur, promoting culturally responsive, play-based, and embodied methods that effectively leverage contextual environments. The critical framing, though underdeveloped, directly engages the sociotechnical question of 'who benefits,' seeking to dismantle systemic oppression, bridge digital divides, and protect marginalized populations. To resolve inherent tensions across these diverse dimensions, the review authors propose adopting a posthumanist theoretical lens. This philosophical perspective critiques anthropocentric assumptions, viewing artificial intelligence not merely as an instrument for human utility, but as an agential material entity engaged in intra-action within broader social and ecological networks. This approach aligns with Rosi Braidotti’s affirmative ethics of mutual interdependence and ecological frameworks that position artificial intelligence within a context of planetary coexistence. Finally, the critical review highlighted substantial geographic imbalances in the existing literature. It noted that research is overwhelmingly centered on Western contexts, leaving noticeable empirical gaps regarding early childhood traditions and AI integration across the Global South. Analyzing research conducted in Asian settings, scholars such as Jing Xu identified that regional studies frequently remain broad without achieving sufficient empirical depth. Furthermore, pedagogical traditions, such as East Asian Confucian teaching models, which emphasize teacher-directed authority, present distinctive adaptation challenges for child-centered artificial intelligence applications. However, the authors noted that posthumanist ethics and an ethic of ecological care naturally resonate with long-standing Asian philosophies, including Buddhist and Daoist principles that prioritize universal harmony and relational interdependence among living and non-living entities. For early childhood practitioners, the review offers crucial practical implications: it recommends adopting developmentally tailored AI tools that match cognitive capacities, utilizing multimodal robots to foster collaborative social expression, maintaining ethical transparency regarding machines' psychological limits, and ensuring equitable resource distribution to prepare young children for an increasingly automated and interconnected society.
