English Language Anxiety (ELA) remains a significant psychological barrier for learners in Indonesia, particularly in the context of performance-based assessments. This study investigates the potential of Artificial Intelligence (AI) as a tool to mitigate this anxiety, focusing on student perceptions and the relevance of specific AI features. Employing a quantitative survey design, data were collected from 90 high school students in Bandung using a questionnaire that integrated the Foreign Language Classroom Anxiety Scale (FLCAS) and the Technology Acceptance Model (TAM). Descriptive statistical analysis revealed that students experience high levels of performance anxiety, especially when speaking without preparation. Results indicate a strong positive perception of AI's ability to create a low-stress environment, primarily by reducing the fear of making mistakes (Mean=3.67). Concerning the most relevant AI features, the capability for L1 scaffolding—explaining concepts in Indonesian—was deemed most crucial (Mean=3.88), highlighting its role in alleviating cognitive anxiety. However, a significant concern regarding the transferability of AI-practiced skills to real human interaction was identified (Mean=3.56), moderating the overall perceived efficacy of AI for anxiety reduction. The study concludes that AI is a highly effective affective tool for safe practice but must be integrated within a hybrid pedagogical model that bridges the gap between AI-assisted learning and authentic social communication to fully address the multifaceted nature of foreign language anxiety.
Sbastian et al. (Wed,) studied this question.