AI-ASSISTED ACADEMIC WRITING: PERCEIVED MOTIVATIONS AND STRATEGIES FOR RESPONSIBLE USE

Abstract

Existing discussions of artificial intelligence (AI) in academic writing often focus on whether its use is acceptable, rather than why students use AI and what practices they believe can support responsible use and academic integrity. This qualitative descriptive study examined the reported perceptions of 84 Accounting undergraduates at a Malaysian public university. Participants responded to two open-ended survey questions concerning their perceived motivations for students’ use of AI in academic writing and how to use it responsibly to maintain academic integrity. Responses were analysed using inductive qualitative content analysis. Non-exclusive descriptive categories were developed because individual responses could express multiple motives or responsible-use practices. Students reported motives regarding generating ideas and initiating writing, improving efficiency and organising tasks, improving language and writing, supporting understanding and clarification, alongside assisting with structure or information. Participants’ conceptions of responsible use included preserving authorship, keeping AI within a bounded support role, limiting its use, verifying its outputs, and following institutional rules. Some participants instead proposed using detector scores, paraphrasing, or other surface-compliance strategies. These findings indicate a perceived boundary between support and substitution: students generally regarded AI as acceptable when it assisted their thinking or expression, but less acceptable when it replaced their intellectual contribution or writing work. Conceptually, this boundary frames responsible AI use as the preservation of human authorship through bounded delegation, critical verification, and transparency. However, the limited emphasis on verification and disclosure, together with reliance on technical detection, identifies specific gaps in students’ responsible AI literacy. The findings represent students’ brief, self-reported perceptions and do not demonstrate their actual AI-use practices, compliance with academic-integrity requirements, or effects on learning.