The ability to write clear and descriptive texts is an essential skill in a variety of academic and professional contexts. From designing compelling narrative essays in literature class to crafting clear and concise reports in the workplace, effective descriptive writing allows individuals to communicate their ideas and engage audiences with vivid imagination and sensory detail. However, the development of these skills often requires intensive practice and constructive feedback. Traditional writing instruction often relies on human feedback from teachers and lecturers. Therefore, this study aims to compare the feedback ability generated by generative artificial intelligence on the ability to write in descriptive texts. This study uses a quantitative research approach with quasi-experimental design. This study involved 58 students of Civil Engineering Vocational Education Program. The non-equivalent control group design was used to compare the results of the experimental class and the control class. Based on the results of data analysis using the Wilcoxon test, there were 29 positive ranks, this means that there were 29 students who experienced an increase in scores at the time of the posttest with an increase of 18.38% from the pre-test score. The average Gain Score = 0.4933 with a maximum value of 0.7272 and a minimum score = 0.28. It can be concluded that the use of generative artificial intelligence in providing feedback on students' descriptive texts is a medium or quite effective.
