AI-assisted feedback in nursing education: A Human-in-the-Loop Assessment Framework for ethical and pedagogically sound practice.

The increasing adoption of generative artificial intelligence (GenAI) in higher education has generated significant opportunities to support assessment and feedback processes. In nursing education, where assessment is closely linked to the development of clinical judgement and safe practice, the integration of AI-assisted marking presents both opportunities and risks. This paper synthesises emerging empirical and policy literature to examine the reliability, pedagogical implications, and ethical challenges of AI-assisted iterative feedback in nursing education. Evidence suggests that while GenAI can provide rubric-aligned feedback and may support efficiency in large-cohort contexts, it remains limited in its ability to capture nuance, support reflective learning, and evaluate complex, context-dependent competencies central to nursing practice. Concerns relating to bias, transparency, and data governance further complicate its implementation. In response, this paper proposes a Human-in-the-Loop Assessment Framework to guide the responsible integration of AI-assisted feedback in nursing education. The framework conceptualises assessment as a multi-layered process comprising structured inputs, AI-generated draft feedback, educator-led review and validation, pedagogically meaningful feedback outputs, and cross-cutting governance and ethical oversight. These stages delineate where AI may appropriately support assessment and where educator review and decision-making are required. The framework advances current discourse by offering a structured, scalable approach to integrating AI into assessment that operationalises human oversight across the assessment process. It provides a practical basis for defining the respective roles of AI and educators and for guiding future research, policy development, and implementation in AI-mediated assessment practices.