Generative Artificial Intelligence (GenAI) has been reshaping creative practices across various domains. This paper focuses on music composition as a case study to explore the nature of human-AI interaction. Prior research has often focused on the technical capabilities of musical AI and the effect of current music AI models on composers. However, this research overlooks the intricate nature of music composition, which involves a complex interplay of personal motivations, socio-cultural factors, and the unique characteristics of different musical genres and styles. To address this gap, after discussing the importance of context and AI roles in human-AI interaction, we propose an ethnographic study that investigates the practice and needs of music composers. Through semi-structured interviews and participant observation we can look into the situated nature of music creation. By doing this, we aim to inform the design of better human-AI interactive systems that support composers’ creative practice and empower them to explore new musical possibilities, rather than supplanting their creative skills with autonomous systems.