The Impact of AI Virtual Tutor ‘DeepSeek’ on Students’ Writing Proficiency

Abstract

This study investigated the impact of the AI virtual tutor DeepSeek on students’ writing proficiency among tenth-grade students of a Madrasah Aliyah in Pamulang, Indonesia, using a quantitative pre-test–post-test design with a sample of 31 students. Students completed five essay-based writing tasks measuring content, organization, vocabulary, language use, and mechanics. Their scores were analyzed using the Wilcoxon Signed Rank Test to determine the significance of improvement after four weeks of DeepSeek-assisted practice. The results showed a statistically significant increase in overall writing performance, indicating that real-time feedback from DeepSeek supported students’ self-regulation, accuracy, and clarity in writing. The novelty of this study lies in examining DeepSeek an emerging AI tutor that has received very limited empirical investigation within an authentic classroom setting, offering new evidence of its pedagogical value compared to more commonly studied tools such as ChatGPT or Quillbot. The study concludes that DeepSeek is a valuable instructional aid and recommends its integration into English writing instruction to complement traditional teaching.