EDUCATIONAL ASSESSMENT IN THE AGE OF AI: A NARRATIVE REVIEW ON DEFINITIONS AND ETHICAL-LEGAL PRINCIPLES FOR TRUSTWORTHY AUTOMATED SYSTEMS

This study presents a narrative review of scholarly literature at the intersection of education, explainable artificial intelligence (XAI), and automated assessment algorithms. Acknowledging the pressing need for a unified regulatory framework to enable the safe deployment of AI-based assessment technologies in educational environments, the paper investigates whether a coherent and consistent terminology is employed across academic research. It further compiles and systematizes the ethical and legal principles underpinning the use of such systems, from the Ethical "AI explainability" to the emergent "Right to Explanation" introduced by Article 22 of the General Data Protection Regulation (GDPR). The review finds that key concepts such as "assessment" and "explainability" are currently used with divergent meanings across disciplines. The analysis identifies transparency, accountability, fairness, privacy, and non-discrimination as the core principles most frequently emphasized in the literature, highlighting the need for interdisciplinary convergence to ensure trustworthy and equitable AI integration in Education.