Team Experience and Human Authorship in Creative Writing with AI: The Role of Artistic Background

Abstract

As generative Artificial Intelligence (AI) becomes increasingly integrated into creative practice, a central question concerns how users experience collaboration and authorship when creating with AI. While many existing systems rely on conversational interfaces, alternative interaction paradigms may shape these perceptions differently. We present results from a field study using the Syntactic Synthesizer, a custom-built instrument-like interface for a Large Language Model (LLM) that enables the creation of poems through keyword input and physical manipulation of model parameters. Deployed at a public arts festival, the system supported an embodied, non-anthropomorphized form of human-AI collaboration. We examine how human-AI team experience relates to creativity support and authorship attribution, and whether this relationship differs between users with and without an artistic background. Across 117 participants, perceiving the Syntactic Synthesizer as a good team partner was positively associated with creativity support. Moreover, artistic background moderated the relationship between human-AI team experience and authorship attribution: among artists, stronger team perception was associated with stronger attribution of human authorship, whereas no such relation was observed among non-artists. The moderation remained robust when controlling for prior AI experience, creative self-efficacy and perceived creativity support. Our preliminary findings suggest that experiencing AI as a collaborative partner is not necessarily in tension with feeling like the author of the work. For HCI research, they also highlight the importance of studying alternative non-conversational interface paradigms that help integrate AI into a human-led creative process.