A recent study revealed that the AI chatbot ChatGPT was found to be no match for human accounting abilities. Microsoft-backed OpenAI has developed GPT-4, a machine learning model that has the capability to generate natural language text. This AI product has significantly impressed researchers by passing the bar exam with a score in the 90th percentile, passing 13 of 15 advanced placement exams, and getting a near-perfect score on the GRE Verbal test.
Despite its remarkable achievements, as Jessica Wood (a freshman at BYU) puts it: “It’s not perfect; you’re not going to be using it for everything”. A research team led by David Wood, a BYU professor of accounting, tested the original version of ChatGPT on accounting exams and the results were eye-opening. On average, students scored higher than the ChatGPT chatbot with an overall average of 76.7% compared to the AI bot’s overall mark of 47.4%.
However, ChatGPT did manage to outscore the students in certain areas, such as AIS and auditing. When it came to questions that determined a student’s knowledge of taxes, finances and management, the AI bot did not perform well, possibly due to difficulty processing the mathematical and numerical components involved.
The study also found that ChatGPT found it harder to answer questions when the difficulty level was upped. The AI bot performed better on true and false questions and multiple-choice topics while it struggled with short-answer questions. Additionally, ChatGPT often wrote accurate descriptions but still chose the wrong multiple-choice option. Ironically, ChatGPT occasionally made mathematical mistakes and incorrectly added or subtracted numbers in a bid to answer the questions.
Microsoft-backed OpenAI is a global leader in artificial intelligence research and development. Founded by some of the world’s top minds in the field of AI, their research focuses on the development of human-like AI agents. Their products, including GPT-4, strive to bridge the gap between machines and humans and thus drive the world into the future with intelligent interaction.
The lead researcher of the BYU study, David Wood, is a passionate accounting professor who has a proven track record of articles published. He has worked at Brigham Young University for over thirty years and has been part of many projects such as the First-Year Experience Program. David believes the latest findings could help improve the efficiency of teaching and the learning process for students.
