Two pathways at once: An investigation of how reactance and ethicality shape student learning with AI

Generative AI tools promise to expand student agency in management education, yet the same tool can help and hurt learning inside a single classroom. We argue this contradiction is the signature of two psychological pathways running in parallel. Drawing on the cognitive-affective-conative framework, we develop a dual-pathway serial mediation model in which a cognitive pathway runs from psychological reactance through autonomy perceptions to effectiveness judgments, while an affective pathway runs from ethical perceptions through emotional engagement to effectiveness judgments. Ninety-seven undergraduate management students participated in two AI-enhanced activities with different chatbot tools across one semester. Both pathways operated as predicted. More surprising was how the cognitive pathway changed across exposures. At first use, reactance traveled through autonomy, consistent with self-determination theory. By the second activity, reactance bypassed autonomy and suppressed effectiveness directly. The affective pathway remained stable across exposures. For management educators, ethical framing before a student's first AI encounter may do more work than redesigning the tools themselves to preserve student choice. As a single-cohort study using two researcher-built chatbots, the results are preliminary and point to hypotheses for multi-site replication rather than population estimates.