<p>This study advances and empirically elaborates the concept of Distributed Pedagogical Agency (DPA) as a framework for understanding how teacherly judgment develops in AI-mediated instructional design. Conducted in an undergraduate design-based course where students created educational videos introducing responsible uses of generative AI for secondary learners, the study examines how engagement with AI technologies shapes emerging pedagogical decision-making. Rather than evaluating AI performance metrics, the analysis focuses on learners’ interpretive labor—how they justified, constrained, and strategically integrated AI suggestions in relation to instructional goals and anticipated learner needs. Drawing on reflexive thematic analysis of longitudinal reflection reports, peer evaluations, and final multimedia artifacts, the findings reconstruct a developmental progression across three layers of DPA: interactional negotiation with AI tools, reflective mediation of AI affordances and limitations, and the consolidation of pedagogical authorship. Across cases, learners moved from using generative AI as a productivity tool to positioning it as a differentiated instructional resource requiring simulated responsiveness, ethical discernment, and responsibility-taking. The study demonstrates that teacherly judgment in AI-rich environments emerges through the coordinated interplay of human intention and technological mediation rather than from either source alone. Theoretically, it reconceptualizes teacher agency as relational and distributed across human–AI systems. Methodologically, it shows how pedagogical development in technology-enhanced settings can be traced through interpretive accounts rather than interaction logs. Practically, the findings suggest that structured AI-mediated learning design can cultivate technological discernment and pedagogical responsibility essential for educators navigating the integration of generative AI in contemporary classrooms.</p>