BRIDGING THE GAP: HOW CHATGPT ENHANCES ASSESSMENT TO CLOSE LEARNING GAPS IN MIDDLE SCHOOL

Abstract

This study investigates the impact of ChatGPT-supported assessments on addressing students’ learning gaps. With the increasing integration of artificial intelligence (AI) in education, understanding its potential benefits in enhancing students’ academic achievement is of great importance. For this purpose, a quasi-experimental design was implemented within a quantitative methodology framework. The study involved seventh-grade students from a public middle school in Turkey during the 2024-2025 academic year. To assess students’ learning levels, seventh-grade assessment questions from the “The Solar System and Beyond” unit, published by the Ministry of National Education (MEB), were used. These multiple-choice questions consisted of four options, with one correct answer, and underwent validity and reliability testing. Prior to publication, the Ministry of National Education (MoNE) sought expert opinions and conducted pilot implementations to analyze item performance. All questions were designed to be appropriate for students’ developmental levels and were explicitly aligned with the "Solar System and Beyond" unit.The experimental group participated in a ChatGPT-supported question-and-answer process during the “The Solar System and Beyond” unit. Instead of providing direct answers, students received step-by-step guidance toward the correct solution and immediate feedback. The results showed that ChatGPT-supported assessments effectively addressed learning gaps, and the experimental group demonstrated statistically significant improvement in learning outcomes. This study provides valuable insights into the potential of ChatGPT-supported chatbots to reduce learning gaps in educational settings and suggests that they can contribute to improving students’ academic performance. Additionally, the study offers new empirical evidence that AI-supported chatbots can actively facilitate learning by supporting students’ reasoning processes rather than merely providing correct answers, thus presenting a novel approach to AI-supported formative assessment in education.

Ähnliche Arbeiten