Pedagogical Agency in the Age of Generative AI: A Narrative Inquiry into Teachers' Practices in Taiwan's All-Out Defense Education

Abstract

With the rapid emergence of generative artificial intelligence (AI) tools such as ChatGPT, teachers must critically adopt and interpret these tools within value-oriented curricula—an area that remains underexplored [1, 2]. In Taiwan, All-Out Defense Education combines policy mandates with civic and ethical instruction, positioning teachers as both knowledge facilitators and moral stewards [3]. Previous research has primarily focused on language learning and general curriculum contexts, lacking empirical evidence regarding AI use and teacher agency in policy-driven courses [4].

This study employed a mixed-methods design, integrating narrative inquiry with a supplementary survey. Five qualified high school and college teachers were purposefully sampled for in-depth interviews, and 36 valid questionnaires were collected to triangulate and support the qualitative findings. Thematic narrative analysis and descriptive statistics were used.

The findings reveal a clear continuum in teachers' AI adoption, ranging from cautious observation to systematic integration, with a consistent principle of "use without overreliance." Teachers demonstrated iterative, practical-evaluative, and projective agency [5]. A lack of institutional support and clear ethical guidelines remains a significant constraint.

This study contributes by bridging an empirical gap on generative AI within value-oriented curricula, enriching agency theory in educational technology, and offering actionable policy and practice recommendations for the ethical integration of AI in policy-sensitive education.