To facilitate learning in higher education, ChatGPT is generally used by students to seek immediate feedback on their school work. The synergic use of feedback from both traditional channel (i.e., teachers) and emerging GenAI (i.e., ChatGPT) has fundamentally changed the way students learn. In this study, based on cognitivism framework and self-regulated learning theory, a research model was developed to test the relationship between feedback (from both teacher and ChatGPT) and learning outcomes. Self-efficacy and affective engagement are important factors in cognitive learning theories, and are proposed to be mediators along the paths. Data are collected among students who took computing related subjects, and used ChatGPT in programming tasks. Based on data analysis with 300 students’ records, feedback from both teacher and ChatGPT are proved to have significant effects on self-efficacy and affective learning, and the two mediators significantly influence learning outcomes.