Generative AI and the Allocation of Epistemic Responsibility in Undergraduate Finance Education

Abstract

Generative artificial intelligence can produce credible financial analyses while leaving students unable to justify the reasoning those analyses contain. This conceptual paper explains when teacher competence in integrating generative AI can support independent financial understanding. Through theory adaptation, it connects technological pedagogical content knowledge with cognitive engagement and cognitive offloading, using epistemic cognition to specify what learners remain responsible for knowing. The central construct is the instructional allocation of epistemic responsibility: the distribution of obligations to select assumptions, justify inferences, verify evidence, and revise conclusions among students, teachers, and AI-supported activities. The paper argues that competence contributes to learning when it creates feasible opportunities for students to exercise these obligations. Assistance can release capacity for conceptual work, but it can also substitute for the very reasoning students need to acquire. Five propositions explain these competing pathways and their dependence on prior knowledge, access to verification, and assessment incentives. A four-configuration typology and finance-specific thought experiments distinguish productive delegation from dependence, including cases in which accurate AI output still undermines opportunities to learn. The contribution is a conditional account of how professional knowledge becomes educationally consequential through task design. It also establishes why assisted product quality, independent understanding, and justified judgment under uncertainty must remain distinct educational outcomes. This paper develops theoretical arguments without collecting or analyzing primary data.