The evolution of atomic models, from Dalton’s indivisible spheres to Schrödinger’s quantum mechanical representations, is a foundational topic in high school chemistry education. The curriculum emphasizes the integration of core scientific competencies, such as evidence-based reasoning and model construction, into the chemistry education (Ministry of Education of the People’s Republic of China, 2020). However, traditional pedagogical methods (e.g., textbook diagrams, and teacher-centered lectures) often struggle to engage students in the contextualization of the scientific discovery process. Students may memorize historical milestones but often fail to grasp the iterative and debate-driven nature of scientific progress. Many groundbreaking scientific innovations have emerged from intense debates and intellectual clashes, such as the Bohr-Einstein debates on quantum mechanics or the rivalry between Newton and Leibniz over calculus. Regrettably, students rarely have the opportunity to engage directly with these historical figures or participate in such debates in the classroom, which hinders the development of creative thinking and critical reasoning skills essential for scientific innovation. Emerging AI-driven tools, particularly chatbots capable of simulating historical figures and debates (Okonkwon & Ade-Ibijola, 2021; Wang & Xue, 2024), offer a novel approach to bridge this gap (Zawacki-Richter et al., 2019; Almasri, 2024).

This study aims to investigate the efficacy of AI-historical simulations in teaching atomic model evolution to Grade 10 students. By leveraging generative AI to recreate dialogues between scientists like Bohr, Rutherford, and Chadwick, we hypothesize that students will develop a deeper conceptual understanding and refined critical thinking skills compared to conventional methods. Prior research supports the role of interactive simulations in STEM education (Kefalis et al., 2025) and generative artificial intelligence in chemistry education (Yuriev et al., 2024). However, few studies have explored AI’s potential to humanize historical scientific discourse.

The theoretical frameworks guiding this study include situated learning theory (Lave & Wenger, 1991), which posits that knowledge is best acquired through authentic, context-rich experiences, and constructive pedagogy, which emphasizes active student participation in meaning-making. By immersing learners in simulated debates about atomic structure, AI tools may transform abstract concepts into relatable narratives.

Grade 10 marks the beginning of high school in China and serves as a critical period for establishing foundational knowledge networks in science. At this stage, students are expected to master the atomic structure model and understand the historical development of atomic theory, including the key debates and milestones that have shaped modern chemistry (Ministry of Education of the People’s Republic of China, 2020). A pilot study was conducted to refine the experimental protocol and assess feasibility. Four Grade 10 students (age 15-16, with 2 female, 2 male) who had comparable prior knowledge of chemistry were randomly assigned to two groups: (a) a traditional instruction group (n=2), where atomic models were taught using blackboard diagrams and teacher-led explanations of Bohr’s postulates; and (b) an AI-intervention group (n=2), where students interacted with an AI-driven chatbot (“Bohr”) developed on the iFLYTEK (XunFei) platform. The chatbot was fine-tuned using digitized transcripts of Bohr’ s lectures and configured to simulate Socratic dialogues (e.g., “How would you explain discrete spectral lines using my orbital model?”). Both groups completed identical 45-minute sessions on atomic theory fundamentals. Post-intervention, semi-structured interviews were conducted to explore participants’ perceptions of instructional clarity, engagement, and conceptual challenges. In the next three months, this study will employ a mixed-methods quasi-experimental design with two Grade 10 chemistry classes (N=60) randomly assigned to an experimental group (AI-historical simulations) and a control group (traditional instruction). Over four weeks, the experimental group will engage with iFLYTEK (XunFei) platform, a mature chatbot trained on primary sources (e.g., Bohr’s manuscripts, Rutherford’s notes) to simulate debates and role-play scenarios (e.g., designing experiments with period-specific constraints). The control group will receive textbook-based lectures and timeline worksheets. Data collection will include: (a) pre- and post-tests (20-item assessment on atomic model concepts), (b) post-intervention questionnaires (5-Likert-scale and open-ended items on engagement and perceived learning), and (c) semi-structured interviews with a stratified random sample of 10 experimental students to explore experiential insights. Quantitative analysis (paired t-tests, ANOVA) will compare test score gains between groups, while qualitative data (thematic coding of interviews and open-ended responses) will identify patterns in historical empathy and critical thinking. Ethical protocols include parental consent, anonymized data, and chatbot response validation by educators to ensure accuracy.

The pilot findings, where students exposed to the AI chatbot “Bohr” exhibited significantly higher engagement and explicitly requested to continue the AI-mediated lessons, align with broader trends in educational technology. This study demonstrates the transformative potential of AI-driven historical simulations in revitalizing science education, particularly for abstract, historically grounded topics like atomic model evolution. AI’s capacity to humanize scientific discourse, contextualize discoveries within debates, and invite students into participatory rather than passive learning roles (Chen et al., 2025) appears critical to fostering sustained interest. In contrast, the muted engagement of the traditional group underscores the limitations of static, teacher-centered methods in cultivating epistemic curiosity or historical empathy. These results support three key implications. First, AI-historical simulations can bridge the gap between content delivery and process-oriented learning, addressing persistent criticisms of science curricula as overly focused on “final form” knowledge (Allchin, 2013). By simulating the uncertainties and controversies behind atomic models (e.g., Bohr’s resistance to quantum leaps), students confront science as a dynamic, socially negotiated endeavor rather than a rigid canon. Second, the pilot’s success with the iFLYTEK (XunFei) platform suggests that even modestly resourced schools can implement such tools, provided chatbots are carefully scaffolded with historically accurate primary sources. Finally, the pedagogical versatility of this approach, rooted in contextualized narrative-building and adaptive dialogue, demonstrates its potential for cross-disciplinary application, from reconstructing historical revolutions in biology (e.g., Darwin-Wallace debates) to simulating ethical dilemmas in literature or political science. However, limitations remain. The small pilot sample (N=6) and short intervention period preclude definitive claims about long-term retention or scalability. Additionally, while AI-group students asked more hypothesis-driven questions, further research is necessary to disentangle whether this reflects deeper conceptual understanding or novelty effects.

Ministry of Education of the People’s Republic of China. (2020). General high school chemistry curriculum standards (2017 edition, revised in 2020). Beijing: People’s Education Press. Okonkwo, C. W., & Ade-Ibijola, A. (2021). Chatbots applications in education: A systematic review. Computers and Education: Artificial Intelligence, 2, 100033. https://doi.org/10.1016/j.caeai.2021.100033 Wang, Y., & Xue, L. (2024). Using AI-driven chatbots to foster Chinese EFL students’ academic engagement: An intervention study. Computers in Human Behavior, 159, 108353. https://doi.org/10.1016/j.chb.2024.108353 Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education – where are the educators? International Journal of Educational Technology in Higher Education, 16(1), 39. https://doi.org/10.1186/s41239-019-0171-0 Almasri, F. (2024). Exploring the impact of artificial intelligence in teaching and learning of science: A systematic review of empirical research. Research in Science Education, 54, 977–997. https://doi.org/10.1007/s11165-024-10176-3 Kefalis, C., Skordoulis, C., & Drigas, A. (2025). Digital Simulations in STEM Education: Insights from Recent Empirical Studies, a Systematic Review. Encyclopedia, 5(1), 10. https://doi.org/10.3390/encyclopedia5010010 Yuriev, E., Wink, D. J., & Holme, T. A. (2024). The dawn of generative artificial intelligence in chemistry education. Journal of Chemical Education, 101(8), 2957–2959. https://doi.org/10.1021/acs.jchemed.4c00836 Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge university press. Chen, A., Jia, J., Li, Y., & Fu, L. (2025). Investigating the Effect of Role-Play Activity With GenAI Agent on EFL Students’ Speaking Performance. Journal of Educational Computing Research, 63(1), 99-125. https://doi.org/10.1177/07356331241299058 Allchin, D. (2013). Teaching the Nature of Science: Perspectives & Resources. SHiPS Education Press.