Authorial voice in human–machine writing with GenAI: plays and counterplays of meaning

Abstract Writing is a means of negotiating and expressing human identity and authorial voice. With the increased availability of GenAI technologies to automate writing processes, a critical examination of the transformed relations between identity and authorial voice is urgently needed. In this paper, we apply a critical walkthrough methodology with the paid, advanced application—ChatGPT 4.5—to explore how authorial voice can be shaped through human-AI interaction. By applying a reflexive model of GenAI-integrated writing, we examine AI-generated texts and the potential alignments or gaps in the resulting written text with respect to authorial intentions. The analysis discusses prompts used to compose writing aligned with authorial voice, to demonstrate how human authors assert their presence, express their values, and align or distance themselves from particular positions available in generated texts. The paper outlines implications for the pedagogical adaptation of GenAI for writing, emphasing the use of GenAI applications for writing that strengthen authorial voice by generating texts that prioritise human agency and consider polyvocal perspectives.