Saetbyeol Moo  Seoung-Hey Paik

Vol. 69,  No. 6, pp. 299-318, Dec.  2025
10.5012/jkcs.2025.69.6.299

Redox reaction Artificial intelligence Machine Learning Decision Tree Teacher education program Metamodeling perceptions

PDF JATS XML

In this study, an educational program utilizing decision trees was developed and applied to improve chemistry teachers' metamodeling perceptions regarding oxidation-reduction reaction models. The participants included 23 chemistry teachers enrolled in a master's program at a comprehensive teacher training university. The program was conducted online over three sessions, each lasting four hours, for a total of 12 hours. The educational program consisted of six stages: concept verification and motivation regarding the electron transfer model, oxidation state change model, and bond type change model (Goodstein model); exploration of AI tool functions using the Orange3 program; generation of concept-based data for each model; concept refinement through AI modeling; AI-based error diagnosis and evaluation; and reflection and sharing. Teachers generated information about chemical reaction equations to create a decision tree classification model, identified the causes of classification errors through the decision tree when machine learning indicated errors, and corrected misconceptions independently. To analyze the effectiveness of the program, changes in teachers' values regarding oxidation-reduction reaction models, model-based judgment capabilities, and perceptions of scientific metamodeling were examined through pre- and post-surveys. The results indicated that teachers' values regarding oxidation-reduction reaction models shifted from a hierarchical perspective to a pluralistic perspective, and they developed a higher-level judgment capability to clearly recognize the scope and limitations of the models. In particular, the perception of scientific metamodeling progressed from the level of objective explanatory tools (Level 2) to the level of exploratory and pluralistic tools (Levels 3 and 4). AI tools were utilized as effective teaching and learning instruments that facilitated teachers' metacognitive reflection. Teachers had positive learning experiences through immediate error identification and visualization, promotion of collaborative discussions, and enhancement of metacognitive reflection. These results suggest that inquiry experiences utilizing AI are effective in deepening teachers' understanding of the nature of science and leading to changes in their practical teaching strategies. Therefore, it is essential to continuously develop and expand professional development programs centered on inquiry experiences where teachers construct and evaluate scientific models themselves using AI tools. This will enable chemistry teachers to deeply understand the nature of models and design lessons that foster the scientific thinking and inquiry skills required by future society.