LLM Chatbots in Academic Writing: Usability and Ethical Concerns Among Bangladeshi University Students

The rise of large language model (LLM) chatbots, such as ChatGPT and Gemini, is transforming academic writing support in resource-constrained higher education systems like Bangladesh’s. This qualitative study explores how 12 Bangladeshi university students experienced and evaluated the usability and ethical implications of these tools during a structured academic writing task. Using moderated one-on-one sessions, think-aloud protocols, and Bangla-language interviews, the analysis identified ChatGPT’s intuitive interface, mobile accessibility, and time efficiency as key strengths. However, participants faced challenges with prompt formulation, repetitive outputs, and Gemini’s less intuitive, search-oriented design. While both tools aided ideation and drafting, concerns emerged about fabricated citations, over-reliance, and diminished critical thinking, raising academic integrity issues. These findings highlight the need for context-sensitive AI literacy, clear institutional policies, and user-centered, transparent HCI design. The study offers a comparative perspective from a Global South context, characterized by limited digital resources and evolving AI adoption, contributing actionable insights for policy, pedagogy, and design.