Artificial intelligence (AI) systems are increasingly integrated into higher education to deliver personalised, timely, and context-specific feedback. This narrative review with systematic search components synthesises evidence from 29 empirical and exploratory studies published between 2020 and 2025, identified through structured searches in several databases. The studies used diverse approaches, including generative AI (e.g., ChatGPT), learning analytics dashboards, chatbots, natural language generation, rule-based detection, and sentiment analysis. Sample sizes ranged from small pilot cohorts to large-scale experiments with more than 1,600 students.Findings indicate that AI-mediated feedback enhances engagement, writing quality, and performance, with several studies reporting measurable improvements. Students frequently valued AI-generated feedback for its immediacy, specificity, and clarity, often rating it as comparable to human input. However, trust varied across cultural and disciplinary contexts, with some students expressing concerns about over-reliance, reduced independence, and fairness of outputs. Hybrid models that combine AI personalisation with human oversight emerged as the most effective practice for balancing scalability with pedagogical depth.This review demonstrates that AI-driven feedback has strong potential to improve learning outcomes and student experiences, but its integration requires careful attention to ethics, transparency, inclusivity, and workload. Limitations include heterogeneity of study designs and the short-term scope of most interventions. Future research should prioritise longitudinal and comparative trials to assess sustained impacts across diverse higher education contexts.
Mirzaei et al. (2025) studied this question.