Generative AI in Financial Mathematics Problem-Solving: Integration of Linguistic and Symbolic Processing Functions
摘要
In this study, we explored the feasibility of using ChatGPT to solve financial mathematics word problems and generate educational graphics. ChatGPT is an AI-based tool on statistical methods. Still, it can also show deductive reasoning capabilities derived from large-scale language model training to solve financial mathematical problems by combining formulas deductively. This capability is further enhanced by incorporating AI for symbolic processing, providing perfect solutions to fundamental word based problems without assistance. Furthermore, ChatGPT draws the deductive reasoning process as a graph, which enables the efficient generation of visual teaching aids for educational purposes. Concerning bond pricing problems in bond mathematics and housing loan problems with fixed interest rates, ChatGPT can completely solve and draw deductive reasoning plan graphs. On the other hand, to solve housing loan problems with variable interest rates, much human help is needed because the change in interest rates makes the problem much harder. In this paper, we report on deductive reasoning plan graphs that ChatGPT produced at the current stage.