This research explores the dependency on and patterns of artificial intelligence (AI) usage in English language learning among public university students in Bangladesh. The study aims to investigate the reasons behind students' reliance on AI tools, the types of information they seek, and the extent of their dependence. A mixed-method approach, combining both quantitative surveys and qualitative in-depth interviews, was employed to gather data from 120 students. The findings reveal that AI tools, particularly Grammarly and Duolingo, are widely used to improve academic writing, grammar, and vocabulary. Students perceive these tools as highly effective for language learning, as they provide instant feedback and make learning more accessible. However, the study also highlights significant concerns regarding the limitations of AI tools, such as their inability to grasp contextual understanding, cultural nuances, and the risk of over-reliance, which may hinder the development of critical thinking and metacognitive skills. Despite these concerns, students recognize the utility of AI tools as a complementary resource rather than a replacement for traditional learning methods. The research suggests that AI tools should be integrated into language learning in a balanced manner, with attention to equity in access and the development of students' higher-order thinking skills. It also emphasizes the need for policy interventions to address the digital divide and ensure that all students, particularly in resource-limited environments, have equitable access to these transformative tools.