AI in Education: A Critical Review of Applications, Bias, and Equity Challenges

Abstract

Artificial intelligence is rapidly transforming higher education, moving from peripheral support to everyday practice in teaching, learning, and assessment. Applications such as learning analytics, intelligent tutoring systems, automated scoring, and generative AI are widely promoted for personalization, scalability, and efficiency. However, evidence shows that these same tools can reproduce bias, compromise fairness, and destabilize pedagogical and ethical norms. Predictive models often privilege students with stronger linguistic or digital resources, tutoring systems struggle with cultural adaptation, automated scoring raises transparency concerns, and generative AI complicates authorship and academic integrity. This paper presents a critical review of 50 peer-reviewed studies published between 2020 and 2025. Rather than providing a systematic mapping, the review synthesizes debates on applications of AI in education and highlights how bias emerges across data, algorithms, social structures, and human–AI interaction. The analysis underscores the tension between opportunity and risk, arguing that AI’s educational legitimacy depends less on efficiency gains than on fairness, accountability, and contextual sensitivity. The review concludes with future directions across technical, pedagogical, policy, and interdisciplinary dimensions, calling for equity-centered and human-centered approaches to AI in education.