<![CDATA[This study examines the implementation of artificial intelligence (AI)-assisted authentic assessment and its effect on the learning growth of elementary school students. The study employed a nonequivalent control-group quasi-experimental design involving 72 fifth-grade students. The experimental group received AI-assisted authentic assessment through performance tasks, inquiry-based projects, and open-ended activities, whereas the control group used conventional assessment methods. The findings indicate that the AI-assisted assessment system operated effectively. The natural language processing (NLP)-based system mapped students’ levels of conceptual mastery, identified dominant errors, and generated diagnostic information that teachers used to provide rapid, specific, and adaptive feedback. The system also accelerated response analysis, monitored students’ individual understanding, and promoted active cognitive engagement, enabling students to explain, connect, and reflect on their observational findings related to scientific concepts. The results further show that the experimental group achieved a high N-Gain score (0.72), whereas the control group obtained a moderate N-Gain score (0.43), with t (70) = 11.704, p < 0.001, and Cohen’s d = 2.76, indicating a very large practical effect of AI-assisted authentic assessment on students’ learning growth. These findings confirm that AI-assisted authentic assessment is effective in enhancing learning growth, strengthening the formative function of assessment, and supporting more personalized, reflective, and evidence-based science learning.]]>
