A systematic review on the future of educational assessment: AI-driven grading and personalised feedback in higher education

Purpose This review discusses the revolutionary role of artificial intelligence (AI) in educational assessment, specifically AI-powered grading and personalised feedback in higher education. The research outlines how AI improves grading efficiency, consistency, and scalability while minimising biases and human subjectivity. It also assesses the role of AI in delivering adaptive, real-time feedback to facilitate student learning and engagement. While AI provides significant advantages in automating routine evaluation tasks, the review also addresses critical challenges such as data privacy, algorithmic transparency and human oversight for ethical, equitable and effective educational use. Design/methodology/approach The systematic search and selection process followed PRISMA guidelines. This review adopts a systematic and analytical approach in examining AI-based grading and personalised feedback in higher education. It synthesises recent literature, case studies and empirical research to assess the application of AI in automating assessment processes. The study examines various AI technologies, including machine learning, natural language processing and computer vision, to determine their effectiveness in grading and feedback. Secondly, it critically analyses ethical concerns, algorithmic bias and privacy problems associated with data. By bringing together theoretical perspectives and practical applications, this review depicts the benefits and drawbacks of AI-based evaluation. It sheds light on optimum practices for its incorporation into learning environments. Findings The critique highlights that AI-driven grading systems bring about efficiency, scalability and consistency in educational grading with reduced human bias. AI-enabled feedback enriches student learning experiences with personalised and real-time feedback, where adaptative learning is encouraged. Nonetheless, there remain challenges, including ethical concerns, data privacy and algorithmic bias, that should be treated with caution. While AI streamlines grading and makes differentiated instruction easier, human oversight is nonetheless required to ensure fairness and contextual accuracy. The evidence suggests that an AI-human hybrid optimises educational assessment, balancing automation and pedagogical judgement. The function of AI in assessment requires continuous calibration of ethical values and evolving educational needs. Originality/value This review provides an integrated synthesis of AI-driven grading and tailored feedback with a balanced discussion of its strengths and limitations in higher education. It offers new insights by bridging the latest research with practical applications, highlighting the need for a hybrid human-AI approach to balance fairness and contextual appropriateness. The study highlights the transformative potential of AI in testing while tackling issues related to ethics, algorithmic bias and data privacy. Through the presentation of best practices for deploying AI, this review serves as a helpful reference for educators, policymakers and researchers seeking to optimise the role of AI in facilitating student learning and assessment procedures.