The rapid advancement of artificial intelligence (AI) is bringing about changes across society, highlighting the need for competencies to respond to these changes. In science education, such competencies are emphasized in terms of scientific literacy....
The rapid advancement of artificial intelligence (AI) is bringing about changes across society, highlighting the need for competencies to respond to these changes. In science education, such competencies are emphasized in terms of scientific literacy. In particular, the Nature of Science (NOS) has long been addressed as a central topic in science education, where it is recognized as a key component of scientific literacy. However, many students still demonstrate limited understanding of NOS, often perceiving scientific knowledge as objective and absolute. Therefore, educational efforts are required to improve students’ understanding of NOS through appropriate instructional interventions.
Scientific modeling can be considered an effective approach for supporting NOS understanding, as it allows students to directly experience the processes through which scientific knowledge is constructed and developed. For modeling instruction to be effective, teachers need to support students’ modeling processes through feedback and promote metacognition so that modeling activities are recognized as processes of scientific knowledge construction. However, despite acknowledging the importance of modeling instruction, many teachers face practical difficulties in classroom implementation due to time constraints and instructional burden.
In this regard, the potential of generative AI can be considered as a tool to address the challenges teachers face in scientific modeling instruction. Generative AI can provide individualized feedback and support reflection on learning, and thus is expected to enhance the effectiveness of scientific modeling instruction. The purpose of this study was to develop instructional design principles and detailed guidelines for scientific modeling instruction using generative AI to promote students’ understanding of NOS, and to examine their validity and effectiveness.
To this end, a design and development research methodology was applied. Initial design principles and detailed guidelines were derived through a review of prior literature and interviews with teachers, followed by two rounds of expert review to establish internal validity. Based on the refined design principles, a four-session instructional intervention was designed and implemented in a Grade 11 Earth science classroom. External validity and educational effectiveness were examined through pre- and post-tests on NOS understanding, evaluations of student models, and in-depth interviews with teachers and students.
The results indicated overall improvement in students’ understanding of NOS, with statistically significant gains particularly in the domains of tentativeness, subjectivity, and socio-cultural embeddedness of science. In addition, students’ scientific models showed significant improvement across all dimensions, including model construction, representation, and explanatory power. Interview analyses further revealed that generative AI functioned as a teaching–learning partner by supporting students’ cognitive development and understanding of NOS through questioning, feedback, and reflection throughout the modeling process, while enabling teachers to focus more on their roles as facilitators and coordinators of classroom interaction.
Based on these findings, a final set of design principles consisting of six principles and 24 detailed guidelines was developed. The six design principles include the principle of iterative modeling, learner-driven inquiry, metacognitive facilitation, integration of the Nature of Science (NOS), adaptive support, and interaction facilitation. The detailed guidelines were organized by distinguishing supports provided directly by teachers and those delivered through generative AI.
This study is significant in that it systematically developed and empirically validated instructional design principles and detailed guidelines for integrating generative AI into scientific modeling instruction to enhance students’ understanding of NOS, contributing both academic and practical implications.