Akitomo Tatsuya, Hamada Masakazu, Tsuge Yasuko, Kaneki Ami, Ogawa Masashi, Nishimura Taku, Kusaka Satoru, Mitsuhata Chieko, Nomura Ryota
Department of Pediatric Dentistry, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, JPN.
Department of Oral and Maxillofacial Oncology and Surgery, Graduate School of Dentistry, The University of Osaka, Osaka, JPN.
Cureus. 2024 Nov 5;16(11):e73103. doi: 10.7759/cureus.73103. eCollection 2024 Nov.
BACKGROUND/PURPOSE: Artificial intelligence (AI) has developed rapidly and is applied to many fields including dental education. In this study, we evaluated AI performance on the Japanese National Dental Examination.
We extracted 349 of 400 compulsory questions from the National Dental Examinations over the past five years. Questions were presented to ChatGPT 3.5, ChatGPT 4o mini, and Gemini, and their performance was evaluated across 13 topic categories.
ChatGPT 4o mini achieved the passing criteria for exams for two years and had the highest total score of the three AIs. The scores of ChatGPT 4o mini on "Society and dentistry" and "Cardinal signs" were significantly higher than those of ChatGPT 3.5 (<0.05).
The high performance of ChatGPT 4o mini indicates the potential value of the tool in dental education. Further improving its performance may lead to future clinical applications in dentistry.
背景/目的:人工智能(AI)发展迅速,已应用于包括牙科教育在内的许多领域。在本研究中,我们评估了人工智能在日本全国牙科考试中的表现。
我们从过去五年的全国牙科考试中提取了400道必答题中的349道。将这些问题呈现给ChatGPT 3.5、ChatGPT 4o mini和Gemini,并在13个主题类别中评估它们的表现。
ChatGPT 4o mini连续两年达到考试及格标准,并且是三款人工智能中总分最高的。ChatGPT 4o mini在“社会与牙科”和“主要体征”方面的得分显著高于ChatGPT 3.5(<0.05)。
ChatGPT 4o mini的高性能表明了该工具在牙科教育中的潜在价值。进一步提高其性能可能会在未来应用于牙科临床。