ChatGPT의 성능개선 및 검색기능 평가 - 회계분야 문제 해결 사례를 중심으로 - = The Evaluation of ChatGPT’s Improvement and Search Functionality† - Focusing on Accounting Cases -

다국어 초록 (Multilingual Abstract)

[Purpose] OpenAI's ChatGPT was first introduced in late 2022. The company has been released new versions of the model with improvements. In addition, a wide range of attempts has been made to take advantage of generative AI models for specialized fields. This study uses the latest ChatGPT 4o and evaluates its performance, which has been improved since ChatGPT 3.5.
[Methodology] Based on the limitations reported in previous studies using ChatGPT 3.5, this study evaluated whether the performance was improved by utilizing ChatGPT 4o, the latest model as of August 2024. In particular, the performance was evaluated with respect to the accounting area.
[Findings] The OpenAI's ChatGPT 4o correctly solved issues such as cost allocation of inventory assets using the last-in, first-out method, accounting for bond investments using the effective interest rate method, determining cash flows from operating activities using the indirect method, and allocation of overhead expenses using the step method and reciprocal method. However, there were limitations in applying the latest revised tax laws or providing financial information of a specific company with the Internet search.
[Implications] As the performance of general-purpose AI models has improved, the model will be applied to the accounting. It is expected that the application of AI model to various accounting fields will continue, which will open up new possibilities.