Generative artificial intelligence (AI) is increasingly integrated into creative writing, raising questions about its implications for creative outcomes and the experience of creating them. While previous research has predominantly focused on creative products, less attention has been devoted to individuals’ experience of the creative process. Drawing on Self-Determination Theory, research on creative self-beliefs, and human-AI creativity, the present experimental online study examined the effects of generative AI during a creative writing task. Participants (N = 141) were randomly assigned to compose a poem either with ChatGPT assistance or without generative AI support. The study assessed state creative self-perception, perceived autonomy and competence during writing, as well as self-rated creativity and perceived authenticity of the completed poem. Creative self-efficacy in writing was examined as a potential moderator of effects on autonomy and competence. Participants in the AI-assisted condition reported lower creative self-perception, autonomy, and competence than those in the human-only condition. No reliable condition difference emerged in self-rated product creativity, while the difference in perceived product authenticity did not remain significant after correction for multiple testing. Moreover, creative self-efficacy in writing did not moderate the effects of AI assistance on autonomy or competence. Overall, AI assistance was more strongly associated with participants’ subjective experience of the creative process than with evaluations of the resulting work, highlighting the importance of distinguishing between creative process experiences and product evaluations.