AI and Educators in Partnership : Transforming English Language Assessment and Cross-Cultural Competence with a Hybrid Model

The rapid evolution of Artificial Intelligence (AI) necessitates a paradigm shift in English Language Teaching (ELT), expanding educational objectives beyond linguistic competence to encompass critical 21st-century skills. Grounded in Vygotsky's social constructivist framework, this study investigates the efficacy of a hybrid learning model within a Japanese university context. The pedagogical design integrates AI-driven “Exploitation"—utilizing tools such as Progos, CASEC, and various Large Language Models for linguistic scaffolding—with human-mediated “Exploration" via direct interaction with U.S. exchange students. Quantitative analysis of pre- and post-tests revealed statistically significant improvements in speaking range, fluency, and overall proficiency, validating AI's efficacy in fostering foundational skills. Conversely, qualitative data indicated that intrinsic motivation (100% agreement) and worldview transformation (87% agreement) were driven exclusively by human interaction. Notably, only 42.1% of participants perceived AI as effective for developing Intercultural Communicative Competence (CQ), underscoring the technology's limitations in the affective domain. Consequently, this study advocates for a synergistic hybrid model that leverages AI for efficient, data-driven training while reserving human interaction for higher-order affective and cultural growth as the optimal pedagogical strategy for the AI era.