How AI Refines Authorship: Evidence from Text and Content Analysis

Generative Artificial Intelligence (AI) is transforming creative work, yet the mechanisms through which humans and AI co-author narratives remain poorly understood. This study examines how the depth of human–AI interaction shapes authorship and perceived creativity in short-form storytelling. We collected 268 stories written in response to a visual prompt under human-only and AI-assisted conditions, and evaluated them using independent rater assessments alongside corpus-based linguistic analysis. Human-written stories were judged to be more useful—and marginally more novel—than AI-assisted stories overall. However, a deeper human–AI interaction improved creative evaluations relative to minimal prompting, indicating that depth of collaboration is a key mechanism to enable hybrid authorship. Linguistic analyses revealed systematic differences: AI-assisted stories emphasized scenic description and archetypal positivity, while human-written stories showed greater subjectivity, temporal grounding, conflict, and emotional nuance. These findings demonstrate how iterative interaction activates human–AI complementarity, advancing understanding of distributed creativity and informing the design of systems that support richer co-creative processes.