Use of Generative AI in Academic Writing: Student Interaction Patterns and Implications

• Academic writing is a complex process, requiring proficiency, and students are increasingly turning to generative AI (GenAI) tools for support with ideation, drafting, and revision.
• However, little is known about students' interaction patterns with GenAI during academic writing tasks and how these patterns influence writing performance.
• Understanding how student interaction with GenAI differs is essential for designing effective GenAI writing tools and AI-assisted instruction and learning experiences that meet students' diverse needs. 

Students with high AI literacy demonstrated a collaborative approach to engaging with GenAI. They:

• Frequently used metacognitive strategies such as planning, evaluation, and revision. 
• Actively and iteratively refined their prompts.
• Viewed GenAI as a collaborative tool.
• Incorporated GenAI's feedback.
• Achieved significantly higher academic writing performance across all evaluation criteria.

Students with lower AI literacy interacted with GenAI differently. They: 

• Used GenAI less frequently.
• Relied primarily on their own ideas and evaluations.
• Were less likely to revise prompts or integrate feedback from GenAI.
• Engaged less with GenAI throughout the writing process, resulting in more linear and less collaborative interactions.  

Overall, the findings suggest that how students engage with GenAI is more important than whether they use it. Students with higher AI literacy collaborated more effectively with GenAI and produced higher-quality academic writing.