Abstract:

As of 2025, Japan has approximately 800 universities. This figure includes not only national and public universities but also private universities. The number of private universities is about four times that of national and public universities, resulting in an oversupply of higher education institutions. Furthermore, due to Japan's declining birthrate, 59% of private universities are operating below capacity. Consequently, private universities with lower academic standards often find themselves with a significant number of students lacking sufficient motivation for their studies. As a result, a considerable proportion of these students fail to meet the academic requirements for graduation, ultimately leading to rising dropout rates.

To address these challenges, many universities have introduced tutoring programs staffed by senior students. Typically, high-achieving upperclassmen or graduate students are appointed as tutors. However, securing a sufficient number of tutors with the necessary qualifications is difficult in itself. Consequently, at several universities, faculty members often serve concurrently as tutors or mentors.

Identifying students at risk of dropping out early and providing appropriate support is a crucial responsibility for tutors. However, university faculty members generally carry heavy workloads, and taking on tutoring duties in addition to these tasks places a significant burden on them. Ultimately, intervention for students at risk of dropping out must be carried out by the tutoring faculty. If we can support the detection of dropout risk, tutoring effectiveness will improve, enabling the provision of adequate care to students at an earlier stage.

This study proposes a dropout risk detection dashboard designed to support tutors and outlines its key features. Utilizing generative AI for dropout risk detection and tutor recommendation functions, it evaluates multiple indicators such as student learning history and cumulative credits earned. When an increased dropout risk is detected, the dashboard makes it recognizable, enabling early remedial instruction for at-risk students. This proactive approach is expected to address issues at an early stage and contribute to reducing dropout rates.

Keywords:

Generative AI, students dropout detection, tutoring assistance, teacher support.