The use of AI to summarize peer feedback

The School of Life Sciences, in collaboration with the Centre for Educational Development and Support, has begun to develop and implement a bespoke peer-learning, assessment, and feedback tool based on the comparative judgement method of assessment. This tool enables students or staff to compare two student submissions of coursework and decide which they think is the better assignment. During this process, the tool collects a variety of feedback from users. When trialling this software within the Introduction to Physiology and Pharmacology module, around 10,000 comments were collected from a range of students, lecturers, and postdoctoral researchers on 54 student poster submissions. While this data could potentially be a rich source of feedback for all those involved in the module, the sheer number of comments meant there too many to review and digest easily by the marking team. A ‘Lightning AI’ solution was explored, where large language model Application Programming Interfaces (APIs) were used to summarize the comments, and custom reports for each poster were generated. This Lightning AI submission details the challenges and opportunities of such an approach, how the school team currently use the summarized reports, and the planned next steps of allowing students access to this form of summarized feedback. This article was published open access under a CC BY licence: https://creativecommons.org/licenses/by/4.0/ .