Assessment for Learning in the Era of Artificial Intelligence: Rethinking Feedback, Evaluation, and Academic Integrity

Abstract

Artificial Intelligence (AI) technologies have been rapidly expanding, including ChatGPT and other generative AI tools, which have reshaped educational assessment. Assessment for learning (AfL), a focus on formative processes, feedback and student-centred assessment, has become embedded in a changing and increasingly complex technological landscape, with new opportunities and exciting challenges. The purpose of this systematic review is to review and to summarize current research on the field of how AI is changing assessment for learning, especially in relation to feedback practices, evaluation methods and concerns about academic integrity in education. A thorough search was performed on several electronic databases (Scopus, Web of Science, ERIC, and PsycINFO) based on the PRISMA 2020 guidelines. Studies included if they explored applications of AI (generative AI, automated feedback, learning analytics, or adaptive assessment) in assessment for learning in grades K-12 or undergraduate and graduate education. Studies published between 2019 and 2025, 60 peer-reviewed empirical studies and systematic reviews, and well-cited conceptual studies were included. The synthesis indicates that AI technologies are being adopted in a variety of ways in assessments, including automated feedback, adaptive testing, intelligent tutoring systems, and predictive analytics. Research indicates that AI-driven feedback can be comparable to human feedback in some scenarios and enhance learning while some concerns have been raised about algorithmic bias, learning dependency, low agency of students, and new challenges in academic integrity of generative AI. Finding a balance between technology and pedagogy, between AI and humans, between assessment for learning and learning itself is essential to ensure the integration is balanced, human-centred, and maintains pedagogical integrity and teacher mediation, and fosters authentic learning experiences. Designing new assessments, establishing ethical guidelines, and implementing comprehensive AI literacy initiatives are critical to responsible implementation.