<![CDATA[Artificial Intelligence (AI)-based learning evaluation is efficient but lacks nuance and risks bias, and the integration of lecturer assessments into the system remains unclear. This study aims to systematically review Human-in-the-Loop (HITL) models to map lecturer roles and measure the impact of their interventions. The study used a literature review of Google Scholar, IEEE, ACM Digital Library, Scopus, dan ERIC databases (2016–2025) with empirical inclusion criteria; 15 studies were analyzed. The results show that while AI improves evaluation efficiency, three lecturer roles initiator, supervisor, and facilitator generally do not directly improve model accuracy. Conversely, the corrector role, which utilizes lecturer feedback for retraining, has the greatest potential for accuracy improvement, but empirical evidence remains limited. Therefore, a shift from simply “Human-in-the-Loop” to a structured feedback mechanism based on Intelligence Augmentation that enables lecturers to contribute to the continuous improvement of Artificial Intelligence models is needed.]]>
Copyrights © 2025
