Human-AI Collaboration in Academic Writing at the Tertiary Level: Practices, Benefits, Challenges, and Responsible Use

Abstract

Generative artificial intelligence is increasingly reshaping academic writing, requiring greater attention to effective and responsible Human–AI collaboration. This study examined Human–AI collaboration practices, perceived benefits and challenges, academic writing effectiveness, and responsible AI use among 300 tertiary-level students in Karachi, Pakistan. The quantitative cross-sectional descriptive-correlational survey design was used, and descriptive statistics, Pearson correlation and multiple regression were used for data analysis. Human–AI collaboration was moderately high for students in relation to language improvement, idea generation and revision. The main advantages were the enhancement in language, time efficiency, and immediate writing support, and the main issues were overdependence, inaccurate information, fabricated references, and academic integrity. Human–AI collaboration was positively associated with perceived academic writing effectiveness (r = .62, p < .001), while responsible AI use also contributed significantly to writing effectiveness. The results highlight the importance of balanced use of AI in writing that maintains students' critical thinking, intellectual ownership, and human agency and responsibility for learning outcomes.