This study empirically examines whether automated writing evaluation(AWE) systems meaningfully change human teachers’ writing feedback patterns in essay-type assessment. Seventy high school students completed essays and revised them using KEEwi, a Korean-language AWE program, and six initial and six AWE-revised drafts were selected through purposive sampling and were scored and given feedback by 23 Korean language teachers. The results showed that AWE-revised drafts received significantly higher scores than initial drafts, and teachers reported reduced cognitive load when providing feedback. The proportion of lower-level feedback, such as feedback on spelling, decreased significantly, whereas the proportion of higher-level feedback, such as feedback on ideas and elaboration, increased markedly. This study has implications for integrating AWE systems into teaching and learning processes.