Beyond Operational Skills: Teachers’ AI Knowledge and Interaction with Generative AI in Lesson Planning

• Teachers’ AI-related professional knowledge analysed using I-TPACK • Interaction logs reveal how teachers collaborate with generative AI • Teacher–AI interaction shows varying degrees of epistemic agency • AI-related teacher knowledge emerges through situated collaboration • Integrating knowledge frameworks with interaction analysis reveals practice Generative artificial intelligence is increasingly entering educational practice, raising new questions about the knowledge teachers require to engage with AI systems in pedagogically meaningful ways. While research on AI literacy and teacher knowledge frameworks has expanded, most studies focus on teachers’ self-reported competencies rather than how professional knowledge is enacted in practice. This study investigates how K–12 teachers conceptualise and enact AI-related professional knowledge during GenAI-supported lesson planning, with particular attention to the relationship between articulated knowledge and enacted practice. Data were collected through a pre-workshop questionnaire (N = 61), interaction logs documenting teachers’ lesson planning with a GenAI system (approximately 1,300 prompt–response pairs), and group-based SWOT reflections following the activity. Qualitative content analysis was used to analyse questionnaire and SWOT data using the Intelligent TPACK (I-TPACK) framework, while interaction logs were analysed using a collaborative problem-solving framework to examine patterns of teacher–AI interaction. The findings show that teachers primarily articulate AI-related competence in technological and pedagogical terms, while ethical considerations appear less frequently. Interaction analysis reveals that teachers most often position GenAI as the primary generator of instructional content, with fewer instances of iterative negotiation or co-construction. The results indicate a complex relationship between teachers’ articulated knowledge and their enacted practices during interaction with GenAI. By triangulating across self-reported, interactional, and reflective data, the study demonstrates how AI-related teacher knowledge is not only conceptualised as competence but enacted through situated teacher–AI collaboration, contributing to ongoing discussions on AI literacy and teacher knowledge in educational contexts.