<![CDATA[The implementation of deep learning and the Love-Based Curriculum has increased the complexity of lesson planning for Madrasah Ibtidaiyah teachers, many of whom continue to experience difficulties in designing contextual, reflective, and value-oriented instructional plans. Although generative artificial intelligence has emerged as a promising tool for teacher professional development, empirical evidence explaining the factors that determine the effectiveness of AI-assisted training remains limited. This study investigates the influence of teachers' instructional needs and training perceptions on the perceived effectiveness of ChatGPT-assisted lesson-planning training. A sequential explanatory mixed-methods design was employed involving 60 Madrasah Ibtidaiyah teachers in North Jakarta. Quantitative data were collected through questionnaires and analyzed using multiple linear regression, while qualitative data were obtained through semi-structured interviews with teachers representing different levels of lesson-planning performance to explain the statistical findings. The results indicate that instructional needs and training perceptions significantly predict perceived training effectiveness (R² = 0.636, p < 0.05), with training perceptions demonstrating the stronger effect. Qualitative findings further reveal that teachers producing higher-quality lesson plans utilized ChatGPT as a reflective cognitive partner to organize instructional ideas and integrate Love-Based Curriculum values, whereas teachers with lower-performing lesson plans encountered challenges related to digital literacy and pedagogical readiness. These findings suggest that AI-assisted professional development is most effective when it addresses teachers' authentic instructional needs while fostering positive perceptions of AI as a pedagogical support tool, thereby strengthening rather than replacing teachers' professional roles.]]>
Copyrights © 2026
