This study explores the pedagogical opportunities and instructional practices that emerge when elementary preservice teachers design science lessons using generative artificial intelligence (GenAI). Drawing on Chiu’s (2024) fourdomain model—Learning, Teaching, Assessment, and Administration—ten third-year pre-service teachers in South Korea participated in a four-week workshop using ChatGPT to design and refine Earth Science lessons aligned with the national curriculum. The participants documented their lesson planning, AI interactions, and reflections, producing qualitative data that were analyzed thematically. Findings show that participants identified various educational possibilities: GenAI supported idea generation and inquiry scaffolding (Learning), helped structure student-centered strategies (Teaching), improved formative assessments and clarified misconceptions (Assessment), and assisted with lesson preparation and time management (Administration). These possibilities translate into specific pedagogical practices, including revising teachercentered approaches to inquiry-based learning, developing scaffolded materials suited to students’ cognitive levels, and reflecting on their evolving roles as science educators. This study suggests that GenAI can act not merely as a tool but also as a catalyst for pedagogical reflection and professional growth. This highlights the need for teacher education programs to foster critical pedagogical reasoning and ethical AI literacy to ensure thoughtful and responsible use of GenAI in science classrooms.

Abstract
1. Introduction
2. Literature Review
    2.1. GenAI and Its Integration into ScienceEducation
    2.2. Teaching and Learning with GenAI:Opportunities and Challenges
3. Methods
    3.1. Research Design
    3.2. Participants
    3.3. Workshop Structure and Procedure
    3.4. Data Collection
    3.5. Data Analysis
4. Results
    4.1. Perceived Instructional Opportunities inGenAI-supported science lesson design
    4.2. Pedagogical Practices Emerging fromLesson Design with GenAI
5. Discussion
6. Conclusion and Implication
References

• Yoon-Sung Choi(Department of Earth Science Education, Pusan National University, Busan 46241, Korea)
• Yoon-Hee Ha(Center for Education Research, Seoul National University, Seoul 08826, Korea) Corresponding author