Abstract
<p>Artificial intelligence (AI) is rapidly redefining the possibilities of language education, yet its pedagogical and assessment-related implications for speaking development remain fragmented across isolated empirical studies. Despite the proliferation of AI-driven applications for pronunciation training, conversational practice, and automated scoring, there is a notable absence of integrative scholarship that systematically connects technological capabilities with learning theory, instructional design, and long-term communicative outcomes. Addressing this gap, the present chapter provides a critical and comprehensive synthesis of current research on AI-enhanced speaking instruction and assessment. It examines how automatic speech recognition, natural language processing, adaptive analytics, and generative dialogue systems influence cognitive development, learner engagement, anxiety reduction, and performance measurement. By situating AI tools within established frameworks of second language acquisition and educational psychology, the chapter moves beyond technological enthusiasm to interrogate issues of validity, bias, sociopragmatic limitation, and ethical responsibility. It argues that while AI demonstrates measurable gains in fluency, pronunciation, and learner autonomy, its effectiveness depends on theoretically grounded, human-centered integration. In doing so, the chapter offers researchers a structured conceptual map and identifies unresolved questions that demand further empirical investigation, particularly regarding reliability, intercultural competence, and sustainable curriculum design.</p>