Bridging or Widening the Gap? Exploring the Determinants of AI Tool Adoption and Its Differential Impacts on Chinese Learning Outcomes among International Students in China

Abstract

The rapid integration of artificial intelligence (AI) tools into language learning practices has raised critical questions about educational equity, particularly among culturally and linguistically diverse student populations. This study investigates the usage patterns, determinants, and differential impacts of AI tools—including machine translation, intelligent writing assistants, speech recognition, and conversational chatbots—among international students learning Chinese in Jiangsu’s higher vocational institutions. Drawing on survey data from 256 respondents and follow-up qualitative interviews with 18 purposefully selected participants, the study examines how factors such as Chinese proficiency (HSK level), prior technology experience, academic motivation, and cultural background influence both the frequency and the effectiveness of AI tool use. Results indicate significant intra-group variation: while most students use AI tools primarily for vocabulary acquisition and homework completion, a notable subset relies heavily on AI for tasks requiring higher-order thinking, leading to a performance paradox—short-term gains in task completion versus long-term skill erosion. Using hierarchical regression analysis and thematic coding, the study identifies digital literacy (β = 0.215, p < 0.001) and teacher guidance (β = 0.248, p < 0.001) as the most critical buffers against the negative effects of AI over-reliance. Chinese proficiency emerged as the strongest protective factor against divide effects (β = -0.218, p < 0.001). The findings contribute to the theoretical extension of the Digital Divide framework in the context of AI-enhanced language education and offer practical strategies for designing AI-literacy training and usage guidelines tailored to international students in vocational college settings.