Evaluating GenAI-produced feedback on undergraduate bioscience essays against good higher education feedback practice.

The emergence of generative artificial intelligence (GenAI) has led to substantial interest in its potential to support student learning and assessment in higher education. In particular, GenAI's ability to provide immediate and iterative formative feedback on the quality of students' work has substantial potential to enhance student learning and attainment. However, for feedback to be impactful and useful, the quality and utility of feedback generated by GenAI needs to accurately reflect the quality of the work and align with principles for good feedback provision. This study evaluated the feedback that an undergraduate student might obtain from a single, unscaffolded interaction with a GenAI platform (ChatGPT4o), without tutor input, applied to 30 Year 1 Bioscience essays. Using a bespoke rubric for feedback quality alongside qualitative and quantitative analyses, the GenAI-generated feedback was evaluated for alignment to established hallmarks of good feedback practice. Overall, GenAI-authored feedback had useful elements but was limited in scope. While feedback tended to be accurate, specific to the essay, with a strong focus on content and substantial elements of feed-forward advice, critical guidance and motivational feedback were limited. Within the feedback test, neutral observations focused mostly on structure and use of figures. Positive feedback focused mainly on use of language, and areas for improvement mostly on structure, missing content, and citations/references. These findings suggest that GenAI may have some potential to provide instant, formative guidance to students, but its use may require substantial scaffolding by educators in order to be effective.