Undergraduate Students’ Perceptions and Use of Generative AI in Assignment Writing: A Qualitative Study in Sri Lankan State Universities.

Abstract

The rapid availability of Generative Artificial Intelligence (GenAI) tools has created both learning opportunities and academic-integrity concerns in higher education. Although Sri Lankan research has begun to examine university students’ acceptance and perceived usefulness of ChatGPT and other AI tools, less is known about how undergraduates experience and use GenAI specifically in assignment writing, including the ethical boundaries they perceive in practice. This qualitative study explored: (1) undergraduate students’ perceptions of using GenAI for assignment writing; (2) the purposes, patterns, and circumstances in which they use these tools; and (3) the ethical concerns and perceived implications associated with such use. Semi-structured interviews were conducted with ten purposively selected undergraduates aged 21–24 years from four Sri Lankan state universities and five degree programmes. Interviews were audio-recorded with consent, transcribed, and analysed inductively using Braun and Clarke’s six-phase thematic-analysis procedure. Four themes were generated: (1) GenAI as a practical support for understanding, brainstorming, grammar, and organisation; (2) ethical uncertainty and academic-integrity concerns; (3) strategic use of GenAI across assignment stages; and (4) perceived risks of over-relian1ce for critical thinking and originality. Participants generally valued GenAI as a supportive learning tool rather than a substitute for independent academic work. However, uncertainty about acceptable use shaped cautious and sometimes inconsistent practices. The study contributes context-sensitive qualitative evidence on how a small group of Sri Lankan undergraduates negotiate the benefits and risks of GenAI in assignment writing. It recommends assessment-specific institutional guidance, AI-literacy development, and assessment practices that support transparent and responsible GenAI use.