<![CDATA[The transformation of pre-service science teachers’ conceptual understanding into effective instructional practice remains a persistent challenge in science teacher education, widely known as the theory–practice gap. This study examines how AI-supported learning facilitates this transformation among pre-service science teachers. Grounded in the AI-CPT model, a one-group pretest–posttest design was employed. Data were collected through conceptual tests, PCK assessments, microteaching observations, and reflective journals. The results show significant improvements in conceptual understanding alongside substantial reductions in misconceptions, indicating that AI supports deep conceptual change. These gains are closely linked to the development of PCK, as participants demonstrated enhanced ability to represent concepts, design instruction, and anticipate student misconceptions. Furthermore, the findings reveal that this integrated knowledge is effectively enacted in teaching practice, as reflected in strong microteaching performance. Qualitative evidence also highlights the role of AI in supporting reflective and iterative learning through continuous feedback. Overall, the findings support the AI-CPT model by demonstrating a coherent and recursive process linking conceptual change, pedagogical development, and instructional enactment. This study contributes to a process-oriented understanding of teacher learning and highlights the potential of AI as a cognitive–pedagogical partner in science teacher education]]>
