Beyond AI Share: A Preregistered Survey and Vignette Study of Perceived Control, Authorship, and Authenticity in AI-Assisted Creative Practice

AI involvement in creative work can be described by quantity, but quantity alone does not reveal how decisions are organized or whether human input remains consequential. This preregistered cross-sectional online survey with two within-person vignettes examined perceived authorship, authenticity, and control in AI-assisted creative practice. The full sample comprised 429 adults, including 164 creators who used AI. Among these creators, declared AI share was not negatively associated with perceived authorship (r = .141, p = .071), contrary to H1. Perceived control was positively associated with authorship (r = .234, p = .003), supporting H2. Control did not moderate the association between AI share and authenticity (interaction b = .043, p = .559), so H3 was not supported. In the full sample, expressive orientation toward art was associated with creative identity threat (r = .341, p < .001), cautiously supporting H4. The strongest result was the vignette contrast: a human-directed process involving idea formation, selection, and substantial revision was evaluated more favorably than acceptance of a near-final AI output (mean difference = .836, t = 13.628, p < .001, dz = .659). Several short indicators had low reliability, limiting strong inference. Meaningful human control is therefore used only as an interpretive lens, not as a validated model. The findings suggest that digital creative practice and tool design should attend to process structure and consequential human control rather than relying on AI-share estimates alone. This is a preprint and has not undergone formal peer review.