Generative AI ‐Supported Creative Writing Among Primary School Students: A Study of Creative Writing Performance, Writing Self‐Efficacy, and Flow State

Abstract

The integration of Generative Artificial Intelligence (GenAI) into creative writing is a growing trend. However, few studies have examined primary school students' interactions with GenAI alongside their writing experiences and resulting writing products. This study compared creative writing performance, writing self‐efficacy, and flow state across GenAI‐supported, peer‐collaborative and individual writing conditions. It also descriptively examined patterns of student–AI interaction during the writing activity. An experiment involved 78 primary school students in China assigned to three conditions: GenAI‐supported writing ( n  = 20), peer collaboration ( n  = 38; 19 pairs), and individual writing ( n  = 20). Students in the GenAI‐supported writing group obtained higher creative writing performance scores than those in the peer‐collaborative and individual writing groups. Interaction analysis showed that students mainly used GenAI to generate and elaborate ideas. Among episodes with observable uptake, adapted or selective incorporation was more common than direct incorporation, and no complete GenAI response was reproduced verbatim. In addition, students in the GenAI‐supported writing group also reported higher writing self‐efficacy than those in the individual writing group, although the difference from the peer‐collaborative writing group was not significant. However, they reported the lowest flow state. Writing self‐efficacy was positively associated with creative writing performance, whereas no significant association between flow state and creative writing performance was observed in this sample. These findings suggest that GenAI may serve as a resource for developing ideas in primary school creative writing.