Prompting as Practice: Reimagining Creative Writing and Authorial Identity After Generative AI

Abstract

The blistering growth of generative AI has caused devastating existential fears to erupt in creative writing classrooms and threatened to upend centuries-old notions of authorial agency, originality, and text ownership. This study argues that while generative LLMs may seem like a creative erasure, there is a need to shift the paradigm and regard prompt engineering as a legitimate, reflective, and iterative form of literature. Using an empirical pre-post mixed-methods research design, we followed the creative writing students at a university for one semester to examine the impact of using AI platforms on their writing practices and sense of authorship. The quantitative measures for creative self-efficacy and creative ownership were filled out using forced 4-point Likert scales. They were mapped alongside the qualitative prompt diary and structural revision log. Longitudinal dependent t-tests and advanced correspondence analysis biplots showed that a cognitive shift was detected, where initial interactions involving automated text generation resulted in feelings of alienation and creative detachment, while extended iterative prompting practice was associated with a high level of recovery of authorial agency. The statistical outputs visually illustrate this: the writer's identity isn't erased, but rather the writer is repositioned, from mechanical assembly of words to curation of algorithmic outputs, from architectural directing to stylistic refinement. The findings confront higher education with a question to consider beyond the rhetoric of academic honesty versus dishonesty: the idea that fostering academic honesty is a means of shrinking the scope of human authorship is an insidious and dangerous proposition.