The Effect of Prompting Type on the Accuracy of Artificial Intelligence in Solving Practical Problems Involving Financial Accounting Standards

Abstract The emergence of Large Language Models (LLMs) and Generative AI has brought about a new transformation in scientific and professional fields, including accounting. The primary objective of this study is to examine the effect of prompting approaches on the accuracy of solving practical problems related to Financial Accounting Standards. This study is applied in terms of its objective and experimental-laboratory in terms of its research design. To this end, ten challenging problems from various topics in financial accounting were selected, and each problem was presented to the ChatGPT AI model using two approaches: a “Simple Prompt” and an “Engineered Prompt.” Subsequently, the model’s responses were compared with the standard textbook answers and evaluated accordingly. The results indicated that the effect of prompting approaches depends on the nature and complexity of the problem. For simple computational problems, both approaches yielded comparable performance. However, for problems involving conceptual complexity, engineered prompts sometimes led to excessive complexity and misled the model, whereas simple prompts produced more accurate answers in certain cases. Furthermore, the AI model demonstrated the ability to self-correct when provided with feedback. The findings of this study can contribute to the effective and informed use of AI tools by accounting students and faculty members. Quantitative results indicated that the mean accuracy score of Simple Prompting (9.1 out of 10) was higher than that of Engineered Prompting (8.4 out of 10), suggesting that greater prompt complexity does not necessarily guarantee higher accuracy.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-12
DOI
https://doi.org/10.5281/zenodo.24160289
Primary Topic
Accounting Education and Careers
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Effect of Prompting Type on the Accuracy of Artificial Intelligence in Solving Practical Problems Involving Financial Accounting Standards

Marjan Hamdollahi
Zenodo (CERN European Organization for Nuclear Research)
Accounting Education and Careers
preprint

The Effect of Prompting Type on the Accuracy of Artificial Intelligence in Solving Practical Problems Involving Financial Accounting Standards

Marjan Hamdollahi
preprint en

Abstract

Abstract The emergence of Large Language Models (LLMs) and Generative AI has brought about a new transformation in scientific and professional fields, including accounting. The primary objective of this study is to examine the effect of prompting approaches on the accuracy of solving practical problems related to Financial Accounting Standards. This study is applied in terms of its objective and experimental-laboratory in terms of its research design. To this end, ten challenging problems from various topics in financial accounting were selected, and each problem was presented to the ChatGPT AI model using two approaches: a “Simple Prompt” and an “Engineered Prompt.” Subsequently, the model’s responses were compared with the standard textbook answers and evaluated accordingly. The results indicated that the effect of prompting approaches depends on the nature and complexity of the problem. For simple computational problems, both approaches yielded comparable performance. However, for problems involving conceptual complexity, engineered prompts sometimes led to excessive complexity and misled the model, whereas simple prompts produced more accurate answers in certain cases. Furthermore, the AI model demonstrated the ability to self-correct when provided with feedback. The findings of this study can contribute to the effective and informed use of AI tools by accounting students and faculty members. Quantitative results indicated that the mean accuracy score of Simple Prompting (9.1 out of 10) was higher than that of Engineered Prompting (8.4 out of 10), suggesting that greater prompt complexity does not necessarily guarantee higher accuracy.

Zenodo (CERN European Organization for Nuclear Research)
Islamic Azad University, Shahr-e-Qods Branch (IR)
Accounting Education and Careers
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon