Ethical challenges of artificial intelligence in education: A systematic literature review on bias, privacy, and academic integrity

Abstract

The emergence of artificial intelligence in education has given rise to huge ethical issues concerning the existence of algorithmic bias, data privacy, and academic honesty when generator AI, large language models, ChatGPT, predictive analytics, and learning analytics services are highly used and implemented in the educational environment. These have fueled controversies about equity, openness, accountability, trust, digital equality, and responsible AI in post-secondary education and wider education. This is a systematic literature review that follows the PRISMA framework to discuss the new ethical issues of AI in education with particular attention given to bias, privacy, and academic dishonesty. The review emphasis on trends in AI ethics, educational technology, adaptive learning, automated assessment, surveillance of students, and policy frameworks. The results indicate that bias by algorithm often supports educational inequity, displaces low-represented students, and promotes assessment, admissions, and predictive decision-making disparities. The problem of privacy is associated with large sums of student data collected, a poor policy of governing data, lack of informed consent, and the growing application of learning analytics systems that rely on surveillance. One of the major concerns is academic integrity because of the prevalence of generative AI in plagiarism, ghostwriting, unauthorized collaboration, and cheating with the assistance of AI. The review also establishes an increasing focus on explainable AI, human-centered AI, ethical governance, privacy-preserving AI, stakeholder engagement, and AI literacy as solutions to responsible innovation.