With the deep integration of AI technologies across all aspects of basic education, teacher AI competency (TAIC) has become a core driver of digital transformation in education. However, existing assessment instruments lack specialized measures tailored to K-12 educational scenarios. Drawing on frameworks and standards such as UNESCO's “AI Competency Framework for Teachers,” this study constructs a three-dimensional integrated framework of “AI foundational competency–AI application competency–AI ethical norms” by analyzing the primary work scenarios of K-12 teachers and typical applications of AI technologies. Within this framework, Application Competency is anchored in three core scenarios: Teaching Practice, Educational Research, and Classroom Management. Through four stages—literature review, expert interviews, expert validation, and pilot testing—exploratory and confirmatory factor analyses based on 18,295 valid responses demonstrate that the final 22-item scale exhibits strong reliability and validity. Overall, AI competency for K-12 teachers (AIC-K12) is at a moderately high level, with significant differences observed across age, teaching experience, school stage, subject area, school location, and partially across gender. Based on these findings, several research implications are proposed. This study provides a robust instrument and empirical evidence for assessing AIC-K12 and supporting professional development, offering important insights for advancing the development of the teaching workforce and improving educational quality in the AI era.
Data science is a field that merges advancements in computer science, statistics, and machine learning. To date, much data science instruction has been through graduate programs or self-directed course offerings. There are therefore fewer opportunities for formal data science education, especially for K-12 educators and learners who have limited exposure to these fields. While some efforts have emerged to introduce data science in pre-college settings, there is little research regarding learning design strategies that support data science for inservice teachers. As such, this project employs a design-based research (DBR) approach to develop accessible and engaging open-access learning resources as a way to reduce barriers to data science instruction across the educational spectrum. Given the gap regarding data science in K-12, our DBR study identifies the following design elements: (a) revised data science content scope, (b) design standardization for interactive learning, (c) reframing data science for case-based learning, (d) artificial intelligence (AI) towards data personalization, and (e) data science coding through goal-based learning. The project also introduces aspects of the curriculum via a teacher professional development, which provides important insights and challenges regarding data science at the middle and high school levels. Implications for learning design, open-educational resources, data science standards, and others are discussed.
This study develops and validates an interpretable transformer-based model for detecting cognitive presence (CP) in MOOC discussions to serve as a diagnostic foundation for instructional design and facilitation. Specifically, we built this model by fine-tuning the RoBERTa architecture with theory-derived structural cues from the Community of Inquiry framework to capture causal, synthesizing, and action-oriented language. Using data from six Social Work MOOCs (4,260 learners; 227,863 sentences), the model classified discourse into four CP phases and a fifth non-CP category. The cue-augmented RoBERTa achieved strong performance (macro F1 = 0.714; κ = 0.666), and SHAP analyses confirmed the model’s internal logic aligns with theory-consistent linguistic markers, ensuring the pedagogical transparency of the results. Learner-level hierarchical regressions revealed that while discussion quantity predicted behavioral persistence, Integration predicted motivation and satisfaction while Resolution predicted perceived knowledge transfer, functioning as diagnostic proxies that participation volume alone does not capture. These findings shift the role of CP analytics from descriptive observation to prescriptive action through a proposed Pedagogical Diagnostic Engineering (PDE) framework. By mapping automated structural signals to specific instructional failure points, this framework offers concrete tactics for real-time scaffolding, such as Socratic prompts and practitioner spotlights, to move learners beyond exploratory monologues toward deeper knowledge construction and professional application.
Learners often rely on visual cues, such as color highlights, to recall important information in instructional texts. Color can serve as an attention-guidance cue but also a memory cue to promote learning. However, research has not yet thoroughly explored how colors can serve both purposes: signaling relevant information within the learning material and serving as memory cues to foster retrieval during testing. Therefore, this study investigated whether different signaling colors in digital learning environments can serve as memory cues to enhance retrieval and improve learning outcomes for secondary students. In a 2 (signaling color: monochromatic vs. polychromatic) × 2 (memory cue presence: during learning vs. during learning and retrieval) between-subjects experiment, 168 participants learned about auroras in a digital learning environment, presented across five texts and seven instructional pictures. The study assessed learning outcomes, cognitive load during learning, and mental load and mental effort during testing, as well as prior knowledge. Results revealed no effects of memory cue presence or signaling color, nor any interactions. Additional Bayes factors largely supported the null hypothesis across all measures. Results are discussed in terms of both color encoding and superficial processing. Color-based signaling does not automatically function as a memory cue, highlighting the need for instructional approaches that promote meaningful cue encoding and consistent reuse across learning and assessment contexts.
With the rapid development of AI, children have become both frequent users and potential co-creators of intelligent technologies. However, research on AI education in primary schools remains limited, with a lack of effective, developmentally appropriate tools for systematically teaching AI concepts, applications, and ethics. To address this gap, this study develops AIgnite, an AI literacy toolkit for upper primary school students in Grades 4–6. The toolkit combines AI knowledge cards with two stages supported by these cards: (i) the Board Game stage and (ii) the AI Storydrawing stage. The two stages create a progressive learning pathway that supports students in building AI knowledge, collaboratively expressing ideas, and reflecting on ethical issues engagingly and interactively. Developed using a child-centered approach, the toolkit underwent multiple rounds of participatory co-creation and iterative prototyping with 30 participants. A user study with another 42 participants demonstrated the toolkit’s effectiveness in conveying AI concepts and applications and in fostering ethical reflection. Additionally, this research concludes by summarizing design insights for developing AI educational tools for children, providing practical guidance and inspiration for future tool development and media innovation in AI education.
