ChatGPT Falls Short in Accounting: Study
A recent study has shown that the AI chatbot ChatGPT is not as skilled as humans when it comes to accounting. The AI, developed by Microsoft-backed OpenAI, uses machine learning to generate natural language text and has been successful in other fields such as passing the bar exam and AP exams.
However, when it came to accounting exams, ChatGPT failed to outperform human students. The study was conducted by researchers from Brigham Young University (BYU) and 186 other universities. ChatGPT scored an average of 47.4%, while students scored an overall average of 76.7%.
ChatGPT performed well on true/false questions and multiple-choice questions, but struggled with short-answer questions, especially those that required mathematical processes. Higher-order questions were also harder for ChatGPT to answer.
The study found that ChatGPT sometimes makes nonsensical errors and can even make up facts. The authors expect that future versions of the AI, such as GPT-4, will improve significantly in this area.
Overall, the researchers concluded that while ChatGPT has potential, it still has work to do in the realm of accounting.
For more news, stay updated with us on Facebook, Twitter, and Instagram.
