Preparing Young Children for an AI Future: What Two Decades of CT Research Tell Us

• In Event: Computational Playgrounds: Foundations and Futures of AI, Play, and Pedagogy in Early Childhood

Abstract

As AI is increasingly integrated into education, there is a conceptual shift—from early emphasis on computational thinking (CT) to broader frameworks of computational (Kafai & Proctor, 2021) and AI literacy (Ng et al, 2022). This transition is especially important in early childhood education (ECE), where questions of developmental appropriateness, equity, and whole-child learning are paramount (Su & Yang, 2022; Su et al., 2023). CT, defined as “the thought processes involved in formulating problems and their solutions so that the solutions are represented in a form that can effectively be carried out by an information-processing agent” (Wing, 2011, p. 1), remains foundational within AI literacy, making it critical to examine the CT literature that has shaped the field’s trajectory to date.
This paper presents a comprehensive review of 115 empirical studies on CT in ECE, published between 2006—when CT was first introduced as a core construct—and 2024, the inflection point where research attention turned to AI literacy. Our review was guided by three broad questions: (1) What are the general characteristics of existing studies on CT in ECE? (2) How is CT defined and operationalized? (3) How is CT promoted and supported in ECE?
Following Alexander’s (2020) guidance for systematic reviews, we employed a three-step process. First, we created a keyword search using 35 seed articles validated by experts. The final query—(“computational thinking” OR “Robotics” OR “Unplugged” OR “Computing Education” OR “Computer Science Education”) AND (“children” OR “kindergarten” OR “early childhood” OR “early years”)—was used across five databases (ERIC, ScienceDirect, SpringerLink, IEEE, ACM), yielding 8,398 results. A first-pass Bayesian classifier trained on abstracts, followed by hand-coding, produced a final corpus of 115 studies. We conducted content analysis of research design, theoretical frameworks, participant demographics, CT definitions and measures, and pedagogical approaches.
Findings reveal significant gaps: limited diversity in study populations; narrow research agendas focused primarily on feasibility and effectiveness; and a narrow framing of CT as a set of cognitive skills without consideration of how such topics would advance the holistic development of young children in their physical, cognitive, social, and emotional domains (NAEYC, 2022). Similarly, arguments for the interdisciplinary or holistic importance of CT tend to focus on the application of CT in new subject areas without consideration of those disciplines’ existing priorities. Situating our findings within both ECE and K–12 computing education (Kafai et al., 2019; Tissenbaum et al., 2021), we propose ways to diversify research agendas and populations, broaden definitions of CT to align with whole-child learning, and expand pedagogical approaches grounded in both ECE and computing education.
As the field pivots toward AI literacy, this review serves as a crucial foundation. CT is not only historically significant but also a core element of AI literacy frameworks. Our analysis informs future research and practice at the intersection of CT and AI literacy, ensuring that efforts to prepare young children for an AI-driven world build upon and advance the lessons of the CT era.

Authors

• X. Christine Wang, University at Buffalo - SUNY

• Chris Proctor, University at Buffalo - SUNY

• Xintian Tu-Shea, University at Buffalo - SUNY

• Grace Yaxin Xing, University at Buffalo - SUNY

• Zhuoyun Cai, University at Buffalo - SUNY