Human–AI Co-Creation Across Scientific, Narrative, and Creative Work: A Reflective Multi-Case Study of Archival Professional Identity

The growing integration of artificial intelligence (AI), particularly large language models and generative systems, has reshaped professional and creative workflows by enabling users to generate, organize, refine, and transform knowledge and expression.However, existing studies often examine AI use within isolated domains, limiting the understanding of how the roles of AI shift across analytical, interpretive, and creative practices.This study addresses this gap through a single-subject reflective multi-case analysis of AI-assisted workflows across three domains: scientific writing, narrative composition, and creative song production.All cases are situated within the shared thematic context of archival professional identity, particularly the expression of "Bangga menjadi arsiparis." The analysis focuses on reconstructed interaction cycles involving human input, prompting, AI-generated output, human evaluation, and either revision, acceptance, rejection, or selection.The findings show that AI assumed three domain-specific roles: an analytical and structural assistant in scientific writing, a narrative structuring agent in personal storytelling, and a creative generator in song production.The study further identifies an interpretive inverse pattern between AI generative autonomy and the intensity of direct human correction or validation.In scientific writing, human control appeared as verification and methodological judgment; in narrative writing, as meaning preservation and authentic voice maintenance; and in song production, as selection, curation, and final approval.Across the three reconstructed cases, these findings suggest that human control did not disappear as AI assumed more generative roles; rather, its form appeared to shift according to task demands, output expectations, and error sensitivity.The study provides a reflective, cross-domain, practice-based framework for understanding adaptive human-AI co-creation and highlights the importance of tasksensitive strategies, human oversight, and contextual judgment in responsible AI use.