Background: In English as a Foreign Language (EFL) contexts, writing self-efficacy a learner’s belief in their capability to produce specific writing tasks is a critical predictor of writing performance and engagement. With the rapid integration of Artificial Intelligence (AI) in education, AI-assisted feedback tools are increasingly competing with traditional teacher feedback.

Objective: This study aims to evaluate and compare the effects of AI-assisted feedback and traditional teacher feedback on the writing self-efficacy of undergraduate EFL learners.

Method: This study uses a simulated dataset created for academic training purposes. A quantitative, quasi-experimental research design was employed with a sample of 120 undergraduate EFL students divided into two groups: an experimental group (n=60) receiving AI-assisted feedback via an advanced grammar and style checker, and a control group (n=60) receiving traditional handwritten teacher feedback over a 10-week period. The Writing Self-Efficacy Scale was administered pre- and post-intervention. Data were analyzed using SPSS v27.

Key Results: Both groups demonstrated significant improvements in writing self-efficacy from pre-test to post-test (p < 0.05). However, the independent samples t-test revealed that the AI-assisted feedback group exhibited significantly higher gains in the "linguistic conventions" subscale, whereas the traditional feedback group showed marginally higher gains in the "ideation and organization" subscale.