AI-GENERATED JAZZ IMPROVISATION: KZN MUSICIANS’ PERCEPTIONS OF AI AS A CREATIVE TOOL

This qualitative study aims to uncover the perceptions of professional jazz musicians in KwaZulu-Natal regarding AI-generated improvisations as a creative tool. It seeks to fill an essential gap in the field of AI music where 93% of AI music studies focus on the Global North, and African viewpoints are non-existent. Guided by the Human Centred AI, Extended Intelligence, and Computational Creativity research paradigms, this research examines emotional expressiveness and trust in AI black boxes using a case study methodology with 8–12 participants. Data collection will be done through a listening task that compares human and AI improvisations on a 5-point emotional expressiveness scale, followed by semi-structured interviews on cognitive, affective, and ethical aspects of trust in such systems. This research contributes to decolonizing AI music research by placing emphasis on the African musical epistemology such as call and response, polyrhythms, and oral transmission. Ethical considerations include Ubuntu values, data sovereignty and HCAI guidelines. In spite of its geographical constraints and small sample size, this research represents the first-ever empirical investigation into the perspective of African jazz musicians regarding AI-generated improvisations and advances epistemic justice in AI music development. Keywords: AI-generated jazz improvisation, KwaZulu-Natal, emotional expressiveness, black box trust, African epistemology