Generative AI Personalized Feedback and College Students’ Learning Agency: An Analytical Framework Based on Self-Determination Theory
TL;DR
C credibility in standard feedback models should be extracted and considered as a new moderator – AI trust; SDT’s three needs work quite differently when the feedback source is a machine rather than a person; and AI trust functions as a gatekeeper that shapes how much feedback quality converts into need satisfaction.
Abstract
The advent of GAI systems such as ChatGPT has resulted in the accessibility of personalized feedback for higher education learners; yet, we understand very little about how such feedback impacts the development of those learners. This paper relies on the SDT theory in the formulation of an analysis framework linking three concepts: feedback quality (objectivity, usefulness, genuineness), need satisfaction (autonomy, competence, relatedness), and learning agency of college learners. Our central claims include that credibility in standard feedback models should be extracted and considered as a new moderator – AI trust; SDT’s three needs work quite differently when the feedback source is a machine rather than a person; and AI trust functions as a gatekeeper that shapes how much feedback quality converts into need satisfaction. The model moves away from examining the effect of AI on learning outcomes towards examining whether AI might—or might not—cultivate genuinely agentic learners.