GenAI-supported co-creation among humanities undergraduates: Offloading, agency, and future task confidence in a stimulus-organism-response framework.

This study examines how humanities undergraduates' perceptions of GenAI interaction features relate to co-creation and future task confidence within a stimulus-organism-response framework. Drawing on survey data from 517 Chinese university students, we conducted observed-variable path analysis and bootstrap mediation analysis. The revised model allowed residual correlations among organism-level states measured in the same survey session and showed good fit (chi-square = 3.56, df = 2, CFI = .999, TLI = .987, RMSEA = .039, SRMR = .008). Perceived GenAI interaction features were positively associated with beneficial offloading, dependent offloading, agency/autonomy, and self-efficacy. Beneficial offloading, agency/autonomy, and self-efficacy were positively associated with both co-creation and future task confidence, whereas dependent offloading was not significantly associated with either outcome after the other organism variables were considered. Bootstrap analyses indicated significant indirect associations through beneficial offloading, agency/autonomy, and self-efficacy, but not through dependent offloading. Harman's single-factor test indicated that the first unrotated factor explained 35.2% of the variance, below the conventional 50% threshold. Sensitivity analyses further examined a co-creation-to-future-confidence path and disaggregated stimulus features. The findings suggest that humanities students' reported GenAI-supported co-creation and future task confidence are linked less to offloading in general than to productive offloading, agency, and confidence-supportive use.