Epistemological foundations and methodological design of a generative AI-based didactic model for English oral proficiency

This scientific essay presents the epistemological and methodological foundations of doctoral research aimed at designing a didactic model mediated by generative artificial intelligence to strengthen oral English proficiency in public secondary school students. The study addresses persistent difficulties in fluency, accuracy, and communicative self-confidence observed in tenth-grade students at the La Paz Educational Institution, located in Apartadó, Antioquia, a context marked by technological limitations and high levels of language anxiety. Methodologically, the research adopts a sequential explanatory mixed-methods approach, articulated with design-based research, which allows for the integration of contextual diagnosis, the pedagogical co-design of the model, its classroom implementation, and its validation through oral performance metrics and qualitative analysis of learning experiences. The critical literature review reveals gaps in the validation of comprehensive didactic models that incorporate artificial intelligence in vulnerable educational contexts, especially in Colombian public education. Consequently, the research seeks to provide theoretical and methodological guidelines for integrating artificial intelligence in an ethical, contextualized, and human-centered manner in foreign language learning.