Abstract:

Formative feedback is widely recognized as one of the most powerful drivers of student learning, yet its consistent and personalized delivery remains a persistent challenge in large first-year university courses. The emergence of Generative AI (GenAI) tools — such as large language model-based assistants — has opened new possibilities for automating, scaling, and personalizing feedback processes. However, the pedagogical effectiveness of such tools depends not only on their technical capabilities, but also on how students perceive, trust, and engage with AI-generated feedback.

This paper presents a theoretical case study that synthesizes existing empirical literature to construct a representative scenario of first-year undergraduate students interacting with GenAI feedback tools in introductory STEM courses. Drawing on research in educational psychology, human-computer interaction, and AI in education published between 2019 and 2025, the study examines three key dimensions of student perception:
(1) perceived usefulness and accuracy of AI-generated feedback,
(2) emotional and motivational responses compared to instructor feedback, and
(3) trust, skepticism, and dependency patterns toward GenAI tools.

Findings from the synthesized literature suggest that while first-year students generally value the immediacy and availability of AI feedback, they frequently question its empathy, contextual sensitivity, and authority. A recurring tension emerges between efficiency and authenticity in feedback interactions. The paper concludes by proposing a set of design principles for integrating GenAI feedback in introductory university courses in ways that are pedagogically sound, emotionally responsive, and critically informed.

Keywords:

Generative AI, formative feedback, student perceptions, first-year students, STEM higher education, AI-generated feedback, human-computer interaction, theoretical case study.