Enterprise AI Analysis

Co-Writing with AI: An Empirical Study of Diverse Academic Writing Workflows

This research provides empirical evidence that students' engagement with AI writing tools is selective, task-dependent, and organized around distinct needs within the academic writing process. Most students concentrated their use within two to three stages of the writing process, rarely applying AI uniformly across all stages. The study identified three recurring configurations: (1) early-stage use (ideation, sourcing, planning); (2) late-stage use (drafting, reviewing); and (3) peripheral use linking early and later stages. These clusters were associated with distinct configurations of perceived benefits and concerns. Learning-related benefits were most salient at the start of the process; quality gains and authorship concerns converged during drafting and reviewing; and productivity benefits appeared at both entry and exit points. Deeper, iterative use concentrated in planning and drafting, and more procedural use in sourcing and reviewing. Overall, AI integration reflects shifting trade-offs between learning, productivity, quality, and authorship across the writing process.