Abstract:

GenAI is reshaping assessment in higher education by enabling new forms of interaction with technology. However, its integration requires pedagogical designs that uphold academic integrity and ensure meaningful learning. This study examines GenAI as an external source (comparator) for self-generated feedback, understood as a metacognitive process in which students revise their work by contrasting it with an external source, such as an exemplar, a rubric, or a GenAI response.

The study addresses two research questions:
(1) What actions do online students undertake when generating self-feedback with GenAI? and
(2) How do they perceive and make sense of GenAI’s role as a comparator?

It was carried out in two courses at the Open University of Catalonia (UOC): Qualitative Methods in Criminological Research, an undergraduate course focused on developing students’ ability to design and report a qualitative research project; and Acquisition and Assessment of Language, a master’s course on evaluating children’s language development. The UOC is a fully online university with over 30 years of experience and a large student population, where assessment is formative and typically organised around progressive assignments supported by written feedback.

The procedure was: students first completed an assignment following the standard assessment procedure of their course. Immediately after submission, they accessed a timed online space where they were randomly assigned one of three comparators (exemplar, rubric, or GenAI. Through a reflective scaffolding tool, students generated self-feedback while revising and improving their assignment. For this research, we conducted an in-depth case study by purposively selecting six students who used GenAI and agreed to participate in a follow-up interview. Through a thematic analysis qualitative approach, we analysed their written reflections when they generated self-feedback using a reflective scaffolding tool, and we also analysed the subsequent interviews to understand how they generated self-feedback using GenAI.

Findings show that GenAI-mediated self-feedback is enacted through a set of recurrent actions, ordered by their frequency. Students primarily use GenAI to identify areas for improvement in their initial assignments, then revisit their understanding of key concepts, followed by detecting aspects they had overlooked, and, less frequently, connecting their revisions to new or previously knowledge. Interviews additionally reveal emergent topics that help to explain how students make use of GenAI as a comparator. These include experimenting with prompting strategies to obtain more relevant feedback; directing corrections purposefully depending on their objectives; questioning GenAI’s reliability and credibility; experiencing uncertainty and tension when interacting with the tool; and showing different levels of GenAI literacy, which influence the depth and productivity of their engagement.

This study offers evidence-based implications for the responsible design of GenAI-mediated self-feedback practices in asynchronous online learning environments. It highlights the importance of transparent scaffolding to position GenIA as a metacognitive tool that strengthens student agency while safeguarding academic integrity.

Keywords:

Feedback, self-feedback, assessment, online education, higher education.