ChatGPT in Education: A Review of Ethical Challenges and Approaches to Enhancing Transparency and Privacy
Ibomoiye Domor Mienye, Theo G. Swart
Abstract
The integration of ChatGPT and large language models (LLMs) into education has created new possibilities for personalized learning, tutoring, and automation of administrative tasks. However, these advancements also present ethical challenges. This paper critically examines the ethical implications of deploying ChatGPT in educational settings, with a focus on data privacy, the opaque nature of AI decision-making, and the risks of biased outputs. To address these issues, we outline actionable approaches, including Explainable AI (XAI) techniques and privacy-preserving strategies, aimed at enabling transparency and protecting student data. We also outline frameworks that support human oversight and governance to maintain trust and accountability in Al-driven educational tools.