教育算法主体概念建构下的分层责任归属机制研究OA

As artificial intelligence(AI)becomes increasingly integrated into educational activities,it has developed a substan-tive influence over key processes such as content generation,learning analytics,and instructional decision-making.Existing theories of education law,which are centered on natural persons and legal persons,are insufficient to capture the multi-actor interactive struc-ture formed among AI systems,developers,operators,and users.This inadequacy results in a lack of clear legal grounds for attributing responsibility when algorithms intervene in teaching practices.This study proposes the concept of the"educational algorithmic sub-ject."Based on the algorithm's functional role in teaching,degree of participation,and risk characteristics,educational algorithmic subjects are classified into three types:passive execution,guidance-and-feedback,and deep intervention.On this basis,a stratified responsibility attribution mechanism is constructed:passive execution algorithms place primary responsibility on users;guidance-and-feedback algorithms achieve multi-actor responsibility allocation through a modified presumption of fault principle combined with a collaborative mechanism of proving;and deep-intervention algorithms are subject to strict liability to address the systemic risks they may generate.By integrating tiered responsibility attribution with institutionalized governance pathways,this study further develops a normative framework that links responsibility allocation with risk governance,providing a reference for AI governance in education and for improving the legal framework of"AI+education"in the context of educational digitalization.