Artificial Intelligence in Education: Opportunities, Challenges, and Implications for the Learning Process

Artificial intelligence (AI) is reshaping education faster than pedagogical theory and institutional policy can absorb it. Although reviews of AI in education (AIEd) are multiplying, most catalogue applications or list risks separately, leaving unclear how opportunities and challenges jointly reconfigure the learning process. This study addresses that gap through a systematic literature review guided by the PRISMA 2020 statement. Records were identified through the Consensus academic search engine using two search strings and supplemented by citation searching, then screened against criteria covering publication between 2021 and 2026, peer-reviewed status, an active DOI, and direct relevance to teaching and learning. Twenty-four studies were retained and analysed through thematic synthesis. The results reveal four clusters of opportunity: personalised and adaptive learning, intelligent tutoring and feedback, AI-supported assessment, and augmentation of teachers' work. These are counterbalanced by five clusters of challenge: academic integrity, algorithmic bias and data privacy, overreliance and the erosion of higher-order thinking, unequal infrastructure and AI literacy, and weak institutional governance. Mapping both clusters onto the stages of the learning process, the study proposes a Learner–Teacher–AI Agency Framework, which holds that AI improves learning only when it redistributes, rather than replaces, human agency. Implications are drawn for pedagogy, assessment redesign, teacher professional development, and policy, particularly in developing-country contexts such as Indonesia.