Generative AI and Academic Self-Confidence: Understanding How AI Tools Influence English Learning Motivation Among Multilingual and International Students

Abstract

Though ample research has been done to examine bias in AI-generated outputs, a significant gap remains in user perceptions of confidence and identity while using generative AI tools. Our research examines the impact of generative AI tools on multilingual and international students’ academic self-confidence and motivation to learn English. While these tools can improve grammar and quality of writing, they may also affect identity, dependency, and authenticity. In this paper, we present three findings from qualitative interviews with multilingual and international students. First, we introduce the concept of a confidence boost versus identity gap: while AI enables students to produce polished, academic English and increases confidence in submitted work, it simultaneously reinforces feelings of inadequacy and dependence. Second, we find that AI use produces a tension between dependency and resistance. Some students rely on AI to reduce cognitive load and overcome barriers to starting tasks, while others actively resist it due to concerns about authenticity and adopting an “AI voice.” Third, we identify split effects on language learning motivation. For some students, AI serves as a model that inspires them to improve, while for others it reduces the motivation to develop their own English proficiency. These findings show that generative AI is not a neutral tool, but one that greatly influences students’ sense of identity and confidence in academic work. Thus, we argue that while AI can help multilingual students feel more confident in English-dominated academic spaces, it can also reinforce dominant linguistic norms and reduce skill development. We discuss recommendations for designing AI systems that emphasize authenticity, learning, and skill development rather than simply optimizing for correctness and efficiency.