Abstract

This study investigates the impact of artificial intelligence tools on English writing skills development among 160 Vietnamese university students. The research examines five key independent variables: type of AI tool (ChatGPT, Grammarly, QuillBot), level of interaction with AI, instruction method, initial English proficiency, and duration of AI use. Using a quantitative approach with pretest/posttest design over a two-month period, the study employed multiple regression analysis to evaluate these factors' influence on writing performance. Results revealed that all five variables significantly contributed to writing skills development, collectively explaining 58.1% of variance. Duration of AI use emerged as the strongest predictor (β = 0.331), followed by instruction method (β = 0.243), initial English proficiency (β = 0.181), level of interaction with AI (β = 0.152), and type of AI tool (β = 0.148). These findings suggest that effective integration of AI tools in writing instruction requires consideration of multiple factors, including sustained engagement, structured pedagogical approaches, learner proficiency levels, and critical interaction with AI feedback. The study offers important implications for educators, educational technology developers, and policymakers regarding AI implementation in language education.