Research on the Companion Learning Function of AI under the Background of Digital Education: Taking Deepseek as an Example

With the deepening implementation of the “double reduction” policy, the digital transformation of after-school homework tutoring has become the focus of attention in the field of education. However, there are still significant deficiencies in the existing research in revealing the mechanism of students’ digital literacy on personalized learning and evaluating the long-term impact of intelligent technology on autonomous learning ability. In this study, 804 students from 8 primary and secondary schools were longitudinally tracked for 4 months using the Deepseek intelligent tutoring system as the research carrier through a mixed research method (comparative experiment method + questionnaire survey method). This paper focuses on the mediating effect of digital literacy in AI-assisted learning and analyzes the influence path of generative technology on cognitive remodeling. The empirical analysis shows that AI plays a positive role in students’ after-school accompanying learning. But at the same time, there are concerns about type accuracy, emotion recognition, thought inertia, and privacy. Therefore, this study recommends optimizing algorithms to leverage complex data, guide interactions, and personalize learning while improving student digital literacy, and providing teacher training and supervision.