Large Language Models (LLMs) have the potential to significantly enhance teaching and learning in higher arts education (HAE). However, HAE institutions have been slow to embrace LLMs’ pedagogical potential. These institutions must provide their staff with guidance, resources, and training on effectively utilizing LLMs in educational contexts while upholding ethical principles. In this chapter, we explore theoretically the implementation of LLMs in supporting research writing in HAE. We argue that LLMs could significantly bolster pedagogical frameworks in HAE-based research education. However, user-end prompting is needed to refine LLMs’ contextual capabilities to support the iterative research writing process. Our exploration envisages a pragmatic pathway for training LLMs as part of AI pedagogy (i.e., applying AI tactfully for educational purposes) and invites a scholarly dialogue towards responsibly harnessing the potential of LLMs in refining the pedagogical approaches in research education.