PRISM: Profiles of AI Use, Creativity, and Authorship in High School Writing

Large language models (LLMs) and AI-assisted writing tools such as ChatGPT, Gemini, and Grammarly are rapidly becoming part of everyday schoolwork. Public debate often assumes that these tools either boost students' creativity or undermine their sense of ownership, yet there is little quantitative evidence about how high school students themselves experience AI-assisted writing. This cross-sectional study used an anonymous online survey of N = 246 students at a large public high school in California to examine how frequency of AI use for school writing relates to two psychological outcomes: perceived general creativity in school writing and sense of authorship over AI-assisted work. Students reported how often they used AI tools for brainstorming, drafting, editing, and overcoming writer's block, and rated their creativity and authorship on 1-5 Likert-type scales. Correlational and regression analyses on brief composite scales showed that more frequent AI use was moderately associated with lower self-reported general creativity but higher perceived authorship, even after adjusting for demographic and contextual covariates. Creativity and authorship were weakly negatively correlated with each other. An exploratory k-means cluster analysis suggested three broad profiles of students combining different levels of AI use, creativity, and authorship. Overall, the findings complicate simple narratives about generative AI in schools: in this sample, heavier AI users feel less creative in their school writing overall yet are more likely to experience AI-assisted work as authentically "theirs."