A study on the dual mechanisms of AI teaching assistants' impact on college students' learning motivation: From the perspective of social cognitive theory

Do AI teaching assistants undermine students' cognitive independence while improving learning outcomes? Based on social cognitive theory, this study systematically addresses this critical question through a 16-week quasi-experimental tracking of 480 college students. The research reveals a dual effect of AI teaching assistants: on one hand, they significantly enhance learning motivation by 22.3% through three major mechanisms—personalized feedback, social presence creation, and adaptive pathways; on the other hand, they lead to a 63% increase in cognitive dependence, a 9.2% decline in critical thinking, and a 54.5% rise in social isolation risk. In response to these alienation risks, the study constructs and validates a three-dimensional intervention model: at the environmental level, implementing blended learning design; at the individual level, conducting metacognitive and critical thinking training; and at the behavioral level, establishing a gradual scaffolding withdrawal mechanism. Intervention experiments show that the comprehensive strategy increases cognitive independence by 37.8%, effectively reversing the dependence trend. The "empowerment-alienation-intervention" theoretical framework constructed by this study provides an actionable risk management solution for AI educational applications and holds significant guiding significance for promoting the healthy development of human-AI collaborative learning ecosystems.