Feedback encounters in doctoral supervision: the role of generative AI chatbots
Lasse X Jensen, Margaret Bearman, David Boud, Flemming Konradsen
Abstract
Effective doctoral supervision is heavily dependent upon dialogic feedback. The rise of generative artificial intelligence chatbots raises the question how interactions with a chatbot align with, or diverge from, authentic feedback practices with supervisors. This self-study from a supervision team employs a novel method of combining rich data from authentic supervisory feedback encounters with simulated chatbot conversations. Supervisor and chatbot responses were analysed around two student dilemmas. The encounters were compared for value – both in terms of perception and impact – as well as formative, temporal and relational dimensions. Our analysis suggests that chatbot feedback sits between an interactive human dialogue, and an additional source of student elicited performance-relevant information. Chatbot feedback encounters highlighted the student’s agency and focussed on the task; supervisor feedback encounters were relational, contextual and developmental. The presence of chatbots underline rather than replace, the need for doctoral supervisors’ engagement with feedback practices that lead to meaningful learning.