Analyzing the Implementation of AI-Based Authentic Assessment Using EAP Talk for Assessing University Students' English Speaking Skills

Abstract

Artificial Intelligence (AI) has the potential to support authentic assessment in English speaking courses by providing automatic scoring and immediate feedback. This study aimed to analyze the implementation of AI-based authentic assessment using EAP Talk in an English speaking course. The study employed a descriptive qualitative approach involving 25 undergraduate students from the Information Systems Study Program at STTI NIIT, Indonesia. EAP Talk was integrated into performance-based speaking activities, and data were collected through classroom observations, AI-generated assessment results, and a student perception questionnaire. The findings revealed that EAP Talk successfully supported authentic assessment by generating automatic scores and feedback on pronunciation, fluency, vocabulary, grammar, and overall speaking performance. The application also received positive responses from students in terms of usefulness, ease of use, and feedback quality. The AI-generated results complemented lecturers' professional judgment, contributing to a more efficient and meaningful speaking assessment process. The study concludes that EAP Talk is a promising supporting tool for implementing AI-based authentic assessment in higher education English speaking courses.