This study statistically compared the differences between Chat Generative Pre-Trained Transformer (ChatGPT) and teacher instruction, comparing the use of ChatGPT with teachers' roles in developing English as a Foreign Language writing. A mixed-method approach was employed. The quantitative component examined learners' writing performance, before and after revisions given by ChatGPT, as well as before and after receiving teacher instruction. Additionally, the study explores the differences between the results in the two groups. The qualitative analysis explored outcomes of using ChatGPT and identified teachers' roles through observational data. A total of 50 English-language learners from a public university were involved as a sample. The results, based on Test of English as a Foreign Language Internet-Based Test (TOEFL-iBT) criteria, revealed that teacher instruction led to higher scores and greater improvement than the use of ChatGPT revisions. Key implications of this study highlight teachers' roles in areas ChatGPT cannot replicate: (a) understanding writing criteria, (b) developing individual critical thinking, and (c) offering ethical guidance. Integrating ChatGPT into writing curricula and AI professional development are recommended.