Purpose This paper aims to investigate how written assessments can be redesigned to foreground human connection within generative artificial intelligence (GenAI)-enabled feedback processes, examining how relational feedback practices operate in artificial intelligence (AI)-augmented environments through two case studies in a large Australian business school. Design/methodology/approach Using an exploratory multiple-case study design, the study draws on focus groups, student surveys and teaching team feedback across two units. Data were analysed deductively through the lenses of relational pedagogy and a relational GenAI integration framework. Findings Students engaged selectively with human and GenAI feedback, combining sources according to need and confidence. Effective integration depended on relational design elements including scaffolding, emotional engagement and co-design. GenAI was most effective as a complement to human judgement within structured, relationally grounded feedback systems. Originality/value This paper provides an empirically grounded account of relational feedback practices in AI-augmented assessment and advances design-based knowledge for intentionally combining human and AI-generated feedback.
Vallis et al. (Tue,) studied this question.