<p>Students are increasingly utilizing AI educational tools in their daily learning, complementing human instructors. Yet, little is known about how and when learning assistant type (Human vs. AI) influences students’ innovation behavior. To better understand these ambiguities, based on self-determination theory and organizational climate theory, the current study constructs a dual pathway model and conducts three online experimental studies to explore the mixed effects of learning assistant type on students’ innovation behavior. Study 1 provides sufficient evidence that AI (vs. Human) learning assistants will both enhance students’ perceived autonomy, thereby promoting their innovation behavior (autonomous pathway), and increase students’ self-esteem threat, thereby inhibiting their innovation behavior (competent pathway). Study 2 shows that high organizational readiness of the college will strengthen the autonomous pathway and buffer the competent pathway. Study 3 further demonstrates that the anthropomorphic design of AI learning assistants is beneficial for enhancing students’ innovation behavior through increased perceived autonomy and decreased self-esteem threat. These findings deepen our understanding of the differences of students’ perceptions and responses towards human and AI learning assistants, and they provide valuable practical implications for the promotion of AI learning assistants.</p>