More of the Same: Rendering of Self in LLM Writing and Feedback Processes – the Crucial Role of Teaching Expertise in an Inclusive Higher Education Future

Large Language Model (LLM) capabilities, such as those provided by ChatGPT, are rapidly emerging as a core capability of educational technologies and productivity tools available to support students and educators in writing and feedback processes. Students can now generate topic ideas, have their sentences and thoughts auto completed, and receive feedback on how well or otherwise they have completed a writing task. We have already entered a new paradigm where students can receive generative writing support and instant feedback at all stages of developing their academic writing.

Our research puts to one side the more obvious issues academic integrity to investigate the relational issues that are likely to arise in writing and feedback processes due to existing resource constraints within higher education delivered at scale. These pressures already limit the potential for students and teachers to develop relationships that allow the professional educator to deliver feedback tailored to an individual’s background, development as a writer, and emerging identity. In this context it may be tempting for administrators and teachers to outsource the role of feedback provider to AI. However, what we already know about the limitations of LLMs and how students are affected by and dismiss impersonal feedback, suggests there are significant risks if we unilaterally replace the teaching professional with LLMs.

Our inquiry will develop a conceptual framework to explore the possible impacts of deploying automated tools in higher education writing and feedback processes without and alongside teaching professionals. The research will engage with poststructuralist theory that is closely linked to the development and expression of self through writing and language. In this way we conceptualise the LLM as a discursive force that could destabilise diverse student identities if unmediated by the care of a teaching professional. The inquiry will generate speculative narrative vignettes to highlight and communicate the risks to students and act as objects to facilitate conceptual thought experiments. The outputs of our inquiry will be to publish anticipatory principles for professional practice related to the pros and cons of using LLMs in writing and feedback processes, and to establish a research agenda to evaluate their use in practice.