When AI Miscalculates: A Lesson in Finance Education

Abstract: Generative AI systems, such as ChatGPT, have swiftly been incorporated into educational settings, offering immediate assistance in several domains, from writing support to conceptual explanation. However, in domains that require numerical precision, such as finance, relying solely on ChatGPT may pose significant risks. This conceptual paper critically explores the limitations of ChatGPT in handling financial calculations where the solutions provided by ChatGPT are difference as compared to the manual calculations. Drawing from recent literature (2020–2025), it highlights risks such as inaccurate outputs, hallucinated references, and diminished critical thinking. A simulation of erroneous financial computation is included to underscore the potential educational consequences. The paper concludes with recommendations to promote AI literacy and encourage the responsible integration of AI in finance education. Learners have to use ChatGPT as a guidance but not to rely fully the answer provided by ChatGPT. Keywords: ChatGPT, Finance Education, Computational Error, AI Literacy, Critical Pedagogy JEL Classification Number: G30, G31