E54 From Design to Enactment: GenAI Integration through SAMR "ask AI to prepare a summary... read the summary and check if it's appropriate... ask AI to write 10 multiple-choice questions with answers" (T8B-R2), moving from passive consumption to active assessment. Another literature teacher created a novel-specific chatbot: "I built a chatbot for students about the novel 'The life before us'... I finally felt some enthusiasm" (T19L-R2). In Assessment contexts, teachers fundamentally reconceptualized evaluation by allowing open AI access during exams, with one teacher designing a matriculation-format exam explicitly instructing: "you're allowed to use the computer... and I'm not blocking the internet" (T10L-R2), transforming assessment from factual recall to critical AI use. These activities illustrate how direct student engagement enables progression toward transformative integration (Viberg et al., 2024; Yang et al., 2024). Redefinition activities (10.5% of total activities) enabled entirely new tasks previously inconceivable without AI technology. Teachers created novel interactive experiences that fundamentally reimagined pedagogical approaches. In Learning contexts, one biology teacher orchestrated a Socratic dialogue in which ChatGPT assumed the role of an expert cardiologist during a classroom lesson. The teacher described how the AI "asked us questions and we answered," noting that this interaction "really helped to break the routine" (T9B-R2). In Teaching contexts, teachers transformed content delivery by using AI to create dialogue videos where virtual characters discussed scientific concepts (T5B-R2). In Assessment contexts, teachers created meta-cognitive evaluation experiences where students compared their peer assessments with AI-generated evaluations (T11C-R2), prompting reflection on judgment itself. These activities exemplify transformative potential where technology enables previously impossible pedagogical approaches, representing the highest integration level that research identifies as relatively rare (Puentedura, 2012; Shamir Inbal et al., 2024). This mapping reveals significant variation in integration levels across pedagogical domains, with Learning activities showing notably higher transformative integration. The next section investigates what factors drive these differential integration patterns. Integration factors: Time vs. Integration Approach Analysis of activities classified according to the SAMR framework (Round 1: N=91, Round 2: N=90) revealed that temporal progression alone does not drive pedagogical transformation. Chi-square analysis showed no statistically significant difference in SAMR level distribution between interview rounds conducted 7-8 months apart (χ² (3) = 5.19, p = .158). Table 2 presents the distribution across rounds. While descriptive patterns suggested shifts, a decrease in Augmentation activities (47→36) and increase in Modification activities (24→38), the overall distribution remained stable, challenging assumptions that sustained practice automatically produces higher-order integration. This finding aligns with research documenting substantial gaps between training participation and classroom implementation (Ding et al., 2024; Kong & Yang, 2024). Table 2. Distribution of SAMR Levels in Round 1 vs. Round 2 SAMR Level Round 1 Round 2 Total Substitution 10 7 17 Augmentation 47 36 83 Modification 24 38 62 The critical differentiating factor emerged as implementation approach. Activities were categorized as Future Planning (conceptual ideas not yet developed), Design of Learning Materials (resource creation), or Enactment with Students (direct classroom application). Chi-square analysis revealed a significant association between integration level and SAMR level (χ² (6) = 53.57, p < .001). Specifically,
RkJQdWJsaXNoZXIy Mjk0MjAwOQ==