Wiboolyasarin W.Kiti T.Wiboolyasarin K.Tiranant P.Jinowat N.Mahidol University2025-08-222025-08-222025-01-01Communications in Computer and Information Science Vol.2551 CCIS (2025) , 3-2018650929https://repository.li.mahidol.ac.th/handle/123456789/111725Effective writing is a critical skill in second language (L2) acquisition, yet low-proficiency learners often face significant challenges, which impede their progress and confidence. This study investigates the impact of AI-driven corrective feedback, specifically utilizing ChatGPT, on improving the writing skills of low-proficiency learners. By addressing challenges such as grammatical inaccuracies, limited vocabulary, and difficulties in structuring coherent texts, AI-driven feedback tools offer scalable, personalized solutions to enhance second language writing instruction. A quasi-experimental design was employed with 19 participants engaged in a 10-week intervention featuring iterative writing tasks and real-time feedback from ChatGPT. The results revealed statistically significant improvements across all five assessed components of writing: grammar, vocabulary, sentence structure, coherence, and task achievement. Sentence structure and coherence exhibited the most substantial gains, highlighting the tool’s ability to help learners construct complex sentences and logically organize ideas. The study underscores the transformative potential of AI-driven feedback to support iterative learning processes, reduce teacher workload, and foster autonomy among low-proficiency learners. These findings contribute valuable insights for educators, researchers, and developers aiming to integrate AI technologies into effective language learning frameworks.MathematicsComputer ScienceConference PaperSCOPUS10.1007/978-3-031-98003-9_12-s2.0-10501302264718650937