The Cost of Algorithmic Assessment: How AI Grading Undermines Academic Motivation and Amplifies AI Reliance Intentions

This research examines the psychological consequences of an AI deployment policy wherein instructors use AI to grade student work while discouraging students from using AI to produce it. Drawing on expectancy-value theory, procedural justice, social exchange, and algorithm aversion frameworks, we experimentally test a psychological model in which algorithmic (vs. human) assessment reduces perceived policy fairness and expected grading accuracy, which in turn lowers prospective students’ appraisals of the course and diminishes their intentions to exert effort. Lower willingness to exert effort, in turn, is associated with increased intentions to rely on AI for task completion against instructional policy. These effects are moderated by prior AI familiarity: individuals with less AI experience perceive algorithmic assessment as more unfair and inaccurate, amplifying negative downstream consequences. The findings contribute to the understanding of how the deployment of AI for assessment by those in authority roles shapes behavioral responses, with implications for the design of assessment policies, AI use guidelines, and instructor practice in higher education.