识别与缓解体育教育中人工智能的伦理风险：一项系统综述

BACKGROUND: Physical education is a key public health setting for promoting physical fitness and lifelong healthy behaviors among children and adolescents. However, ethically inappropriate applications of artificial intelligence in physical education (AIPE) may undermine students' bodily autonomy, educational equity, and the quality of their participation in health-promoting activities. This study aims to systematically identify the ethical risks associated with AIPE, analyze their causes and potential harms, and integrate targeted alleviation strategies.

METHODS: Following the PRISMA guidelines, a systematic search was conducted across six databases: SpringerLink, Web of Science, EBSCOhost, ScienceDirect, Scopus, and CNKI. After multiple rounds of screening, a total of 92 studies published in English or Chinese between January 2016 and May 2026 were included. A hybrid deductive-inductive thematic analysis was employed, with two researchers independently coding the included studies and cross-checking their results to extract and synthesize the types of ethical risks associated with AIPE and the corresponding alleviation strategies.

RESULTS: Ethical risks in the technology dimension include data leakage, privacy infringement, algorithmic bias, and algorithmic limitations. Risks in the physical education dimension include homogeneous physical education teaching, threats to teachers' professional roles and agency, deviation from the goals of physical education, homogenization of student development, alienation in teacher-student relationships, alienation in student-student relationships, lack of humanistic care, value alienation, and academic misconduct. Risks in the body dimension include blurred body boundary, body meaning deconstruction, and body value alienation. Based on the analysis of the types, causes, and potential harms of these risks, this study adopts a stakeholder perspective to develop a systematic set of alleviation strategies encompassing three key dimensions: technological governance, educational regulation, and body protection.

CONCLUSION: Achieving trustworthy AIPE requires balancing technological reliability, appropriateness for physical education, and the preservation of human agency. This study provides a theoretical reference for the ethical governance of AIPE within the field of public health. Future research should strengthen empirical validation, foster international cooperation, and advance research on artificial intelligence ethics.

SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261338255, identifier PROSPERO (CRD420261338255).