AI-Enhanced Academic Speaking Skills: A Qualitative Investigation of Digital Tool Integration in Indonesian EFL University Context

Academic speaking proficiency represents a critical competency for English as a Foreign Language (EFL) learners in higher education. Traditional skill approaches often fail to provide adequate individualized feedback and authentic practice opportunities, particularly in resource-constrained Indonesian university contexts. This qualitative case study explores the use of artificial intelligence (AI) technologies in academic speaking and how they affect the development of oral competency, self-assurance, and independent learning practices in EFL learners. We used a qualitative case study approach to gather information from 15 second-year English education students at a private Indonesian university over the course of six weeks through semi-structured interviews, systematic classroom observations, and artefact analysis. Structured speech initiatives incorporated artificial intelligence (AI) techniques such as ChatGPT, text-to-speech programs,and pronunciation feedback systems. The six-phase theme analysis framework developed by Braun and Clarke, validated by several researchers, was used to analyze the data. The analysis revealed four main themes: increased learner autonomy and self-regulation, improved pronunciation and fluency through instantaneous AI feedback, improved self-confidence through nonjudgmental practice environments, and ongoing difficulties with vocabulary acquisition and the development of metacognitive strategies. AI-assisted speech projects show great promise for improving academic speaking proficiency among EFL learners, especially in the areas of targeted skill improvement and confidence building. However, effective integration necessitates thoughtful educational design that maintains human connection for the development of pragmatic and cultural competence while addressing enduring difficulties in lexical development and metacognitive strategy abilities.