Objective: ChatGPT is an AI-driven language model created by OpenAI that possesses the ability to generate text that closely resembles human language. The objective of this study is to assess the efficacy of ChatGPT in the context of the Plastic Surgery Board Certification examination.
Materials & Methods: The Plastic Surgery Board Certification examinations from 2015 through 2022 were used as a source of questions. The single- and multiple-choice options for each question were imported into ChatGPT 3.5 and 4.0. The rate of true answers to the question was recorded.
Results: A total of 1253 questions were included in the analysis, and ChatGPT accurately answered 516 of them, accounting for 41% of the total. The ChatGPT achieved the maximum score of 55% on the 2017 examination, while it obtained the lowest score of 29% in the 2020 examination. For all 1253 questions, the correct response rate of ChatGPT 3.5 for single- and multiple-choice questions is 48% (431/890) and 23% (85/363), respectively. In addition, the correct response rate of ChatGPT 4.0 for single- and multiple-choice questions is 66% (649/890) and 43% (169/363), respectively, while the correct response rate ranges from 36% to 64% for single-choice questions and 13% to 33% for multiple-choice questions in various years. For the board examination, the correction rate was 41% for ChatGPT 3.5, while 59% for ChatGPT 4.0.
Conclusions: The Plastic Surgery Board Certification Examination was not successfully completed by ChatGPT 3.5 but nearly be passed by ChatGPT 4.0. However, it is important to acknowledge that, in each academic year, additional credits are typically necessary for candidates of the Board Certification Examination to successfully meet the requirements. The utilization of large language models (LLMs) and artificial intelligence in ChatGPT holds significant promise in its capacity to influence the field of examination. Specifically, it can contribute to the clarification of information and the advancement of evidence-based medicine.