Generative Artificial Intelligence in Doctoral Writing within Elite Universities: Usage Patterns and Perceived Benefits and Limitations

Abstract

This study employs a cross-sectional convergent mixed-methods design to examine 298 doctoral students’ use of generative artificial intelligence (AI) in academic writing across STEM, Social Sciences, and Humanities in three elite Ethiopian universities. Quantitative data were analyzed using descriptive and inferential statistics, while qualitative data were thematically analyzed to explore perceived usefulness, ease of use, usage patterns, and contextual experiences of AI adoption. Findings indicate that generative AI is widely used as a strategic tool to enhance linguistic quality, writing efficiency, and textual organization, with adoption driven primarily by task-specific needs rather than disciplinary affiliation. AI merges as an academic support system that enhances writing efficiency and idea development, particularly through linguistic refinement and productivity gains, while simultaneously being approached with critical reflexivity regarding its benefits and limitations. The study concludes that generative AI represents a transformative yet cautiously adopted tool in doctoral writing, where its benefits are balanced against ethical and epistemic concerns. These findings inform pedagogical and policy efforts for responsible AI integration in higher education. However, the study is limited by self-reported data and a cross-sectional design, restricting generalizability and causal inference.

Key words: Generative artificial intelligence; doctoral writing; academic integrity; higher education; elite universities

DOI: 10.7176/JLLL/112-02

Publication date: October 30^th 2026