The impact of generative artificial intelligence on students’ academic writing

Abstract

This dissertation investigates the impact of generative AI tools on university students’ academic writing styles, alongside the cognitive and affective dimensions of the writing process, within the context of English for Academic Purposes (EAP). The study explores students’ writing habits and dependency on AI, addressing concerns linked to AI use, like a decline in independent thinking, original authorial voice, and authentic expression. A qualitiative analysis compares pre and post-AI writing tasks, by applying a corpus-informed analyses of textual choices. In tandem, cognitive and affective dimensions are explored through Think-Aloud Protocols and verbal retrospective interviews to understand students’ attitudes and perceptions of AI use. The results show that AI increases word count and writing fluency while reducing time on task, indicating lower planning-stage cognitive load. However, a higher number of pauses during AI use suggest increased metacognitive evaluation. Textually, AI assisted writing shifts towards a more formal, structured academic voice, marked by reduced lexical diversity and diminshed authorial voice. Thereby, AI reconfigures writing behaviours by increasing writing output and reflective processing, with cognitive demand translating from planning to evaluation. Although students’ confidence appears enhanced, this costs originality and voice. Therefore, the findings reveal the need for a more mindful approach towards AI in academia, as it promotes a standardised framework that limits individual variation.