Artificial Intelligence-Empowering Formative Assessment: Research Progress and Future Directions

Abstract

Formative assessment is crucial for achieving a high standard of education, yet it often lacks personalisation and timeliness. The integration of artificial intelligence (AI) into formative assessment has therefore become a central theme in contemporary digital education research. Based on the Web of Science (WoS) database, this paper employs bibliometric analysis and CiteSpace visualization to examine the literature on generative AI-empowered formative assessment from 2019 to 2025. Annual publication trends, keyword co-occurrences, and thematic clusters are analyzed. The findings reveal a research evolution from”theoretical investigation” to”technological adoption”to”multidisciplinary implementation” with research hotspots focusing on intelligent feedback, large language models, deep learning, and adaptive assessment. Contemporary research, however, still remains constrained by technical dependency, low-quality data, and inherent limitations of foundation models and potential ethical risks. Accordingly, this study proposes a systematic framework comprising four dimensions of object, data, technology, and application to reconstruct learner-centered evaluation paradigms, enable cross-layer coordination, and realize a closed-loop intelligent assessment system. The findings have significant theoretical and practical implications for digital transformation and paradigm reconstruction of educational assessment in the era of AI.