Human–AI co-creation in generative AI video: content characteristics, flow experience, and continued creation intention—a hybrid SEM–ANN study

Abstract

Introduction Grounded in the stimulus–organism–response (S-O-R) theory and flow theory, this study investigates the pathways through which five dimensions of artificial intelligence (AI) video content characteristics—generation quality, emotional resonance, visual novelty, interaction, and narrative coherence—influence university students’ AI video creation willingness (AW), with a particular focus on the mediating role of flow experience. Methods A hybrid approach combining structural equation modeling (SEM) and artificial neural network (ANN) analysis was adopted. Using the Jimeng (Dreamina) AI video generation platform as the research context, questionnaire data were collected from 512 Chinese university students with more than 1 month of usage experience. Results All five AI video content characteristics significantly and positively influenced flow experience, which in turn significantly affected AW. Among these characteristics, visual novelty and narrative coherence exerted the most prominent effects on AW. Bootstrap mediation tests confirmed that flow experience significantly mediated all examined pathways. ANN sensitivity analysis further verified that visual novelty was the primary predictor of flow experience, while flow experience was the core predictor of creation willingness. Discussion This study extends the application boundary of S-O-R theory to generative AI creation contexts and elucidates the psychological mechanism through which external stimuli are transformed into behavioral responses. The findings also provide theoretical grounding and practical guidance for AI video platforms to optimize user experience and enhance users’ willingness to adopt AI video creation.