A Theoretical Patchwork of Metaphors for AI Literacy for Young Children: Insights from a Delphi Study

• In Event: Fostering AI Literacy in Early Childhood: Emerging Theoretical and Empirical Insights

Abstract

As AI becomes increasingly integrated into children’s lives, efforts to develop a framework for AI literacy in early childhood education (ECE) are underway. For example, AI4K-12.org proposes five big ideas, and Ng et al. (2021) identify AI concepts, practices, and attitudes. Similarly, despite the lack of theoretical unity in computational thinking (CT) (Wing, 2006; Wing, 2011; Wang & Proctor, 2022), numerous successful implementations have emerged in the last decades. These successes advocate exploring local, emergent definitions of CT across diverse learning environments (Wilkerson et al., 2020), a call that is especially relevant for AI literacy given its emerging nature and rapid evolution of the technology.
To heed this call, we conducted a Delphi study, a rigorous scientific method that uses iterative group communication to examine and discuss emerging topics, set goals, investigate policies and practices (Beiderbeck et al., 2021; Hsu & Sandford, 2007). We assembled a panel of 30 experts from AI, child development, early education, child-computer interaction, and computing education. Over three rounds of focus group discussions and surveys, we explored the questions of why, what, and how to foster AI literacy for young children.

We examine metaphors from expert panel discussions and present a “theoretical patchwork of metaphors” (Sfard, 1998) for AI literacy in ECE. Guided by works on metaphors and thought (Lakoff & Johnson, 1980; Sfard, 1998, 2012), we argue that metaphors shape thinking and practice by creating new ways of understanding and linking intuition with scientific theory. Metaphors are effective in understanding emergent topics like AI literacy in ECE, serving as sense-making tools and catalysts for new knowledge. Sfard’s (1998) concept of a theoretical patchwork advocates integrating multiple metaphors for context-sensitive, and practically relevant understanding of educational phenomena

Our preliminary analysis reveals intriguing metaphors used by expert panelists. Some likened AI to the topic of “death,” suggesting it should be taught only when necessary or when parents feel comfortable. Others compared AI to “germs,” arguing that children can understand abstract ideas and should explore them. Another perspective likened learning AI to “traditional literacy,” emphasizing considering what children should learn about AI at different developmental stages. The metaphor of “learning to read vs. reading to learn” emerged for age-appropriate teaching strategies. Panelists suggested teaching ethical and responsible use of AI (like a “cellphone and tablet”) and understanding AI’s systems to recognize potential biases (unpacking the “black box”).
These diverse perspectives highlight the complexity of integrating AI literacy into ECE. We propose integrating them into a theoretical patchwork by considering: (1) Complementarity: Balancing caution with discovery by considering parents' input, children's experiences at home, and children's agency to explore; (2) Context-Sensitivity: Adapting these metaphors to fit specific contexts to ensure equity and inclusivity; (3) Interdisciplinary Dialogue: Maintaining insights from interdisciplinary collaboration to foster a holistic approach; and (4). Practical Relevance: Aligning metaphors with the needs and experiences of educators and learners to translate theoretical insights into actionable strategies. By integrating these principles, we can create an adaptable framework for AI literacy in ECE that addresses the diverse needs of stakeholders.

Authors

• X. Christine Wang, University at Buffalo - SUNY

• Chris Proctor, University at Buffalo - SUNY

• Zhuoyun Cai, University at Buffalo - SUNY

• Varun Chetan Bhatt, University at Buffalo - SUNY