Creative Practice Challenging Dominant Generative AI Discourses

Artificial Intelligence (AI) is increasingly used across creative practice and the creative industries. This is especially true of Generative AI (GenAI) tools which anyone can use to create music, text, or images from simple text prompts. Developments of GenAI by major multinational corporations driven by agendas of market dominance and return on investment have led to unethical practices such as training GenAI models on content scraped without permission from intellectual property rights holders. Moreover, the need for enormous training datasets means that commercial GenAI models are inherently biassed towards mainstream styles and require substantial compute and financial resources, further marginalising and excluding creative forms and practices beyond the mainstream. Creative practitioners are taking matters into their own hands. This chapter reflects on creative practices in AI and music and audiovisual arts which challenge dominant discourses on GenAI. For example, communities of artists and musicians are building their own datasets and open repositories of AI models to offer more diverse, inclusive, sustainable, and ethical approaches to GenAI. These offer greater artistic agency than major GenAI models and at the same time allow artists to retain ownership of their practice and creative outputs with GenAI.