Towards responsible artificial intelligence in education: a systematic review on identifying and mitigating ethical risks

Artificial Intelligence in Education (AIED) is becoming increasingly influential in the educational sphere, offering significant benefits and presenting ethical risks. This study fills a crucial gap by systematically classifying and analyzing these risks. Using a combined approach of systematic review and grounded theory coding, ethical risks were categorized into three dimensions: technology, education, and society. In the technology dimension, risks include privacy invasion, data leakage, algorithmic bias, the black box algorithm, and algorithmic error. The education dimension risks involve student homogenized development, homogeneous teaching, teaching profession crisis, deviation from educational goals, alienation of the teacher-student relationship, emotional disruption, and academic misconduct. Risks in the society dimension consist of exacerbating the digital divide, the absence of accountability, and a conflict of interest. Based on an analysis of the types, potential triggers, and hazards associated with these risks, we propose strategies spanning three critical dimensions—technology, education, and society, from the perspectives of stakeholders, to address these ethical risks. This study contributes to a concise and precise analysis of the ethical risks associated with AIED, offering practical solutions for the responsible implementation of AIED.