Pemanfaatan Kecerdasan Buatan (AI) dalam Asesmen Pendidikan: Implikasi Etika, Bias, dan Tantangan Implementasi

Abstract

This study examines the utilization of artificial intelligence (AI) in educational assessment, focusing on the ethical implications, biases, and implementation challenges. A total of 376 articles were identified between 1986 and 2025, with 302 articles found to be relevant. After filtering for publications from 2021 to 2025, the number of relevant articles decreased to 264. Of these, 263 articles were not retracted, 262 articles were not duplicates, and 176 articles were available as open access full-text. After further screening based on inclusion criteria (primary empirical research, excluding systematic reviews, scoping reviews, literature reviews, meta-analyses, narrative reviews, framework papers, qualitative studies, or conceptual papers), only 28 articles met the inclusion criteria. Key findings indicate that while AI can enhance the quality and accuracy of assessments, its effectiveness heavily depends on the implementation context and the quality of methods used. Major challenges related to AI implementation in assessment include algorithmic bias, data privacy issues, and lack of transparency in decision-making processes. The study also discusses potential mitigations for these issues through bias audits, fairness-aware algorithms, and human oversight in AI-based assessment systems. Overall, AI has the potential to improve fairness and accessibility in educational assessment systems, provided it is implemented with inclusive design and careful attention to ethical and bias-related concerns.