Artificial Intelligence in Periodontology: Performance Evaluation of ChatGPT, Claude, and Gemini on the In-service Examination
Authors:
Preprint
Abstract
Background: Artificial intelligence (AI) language models have shown potential as educational tools in healthcare, but their accuracy and reliability in periodontology education require further evaluation. In this study we aimed to assess and compare the performance of three prominent AI language models—ChatGPT-4o, Claude 3 Opus, and Gemini Advanced—with second-year periodontics residents across the United States on the American Academy of Periodontology 2024 in-service examination. Methods: We conducted a cross-sectional study using 331 multiple-choice questions from the 2024 periodontology in-service examination. We evaluated and compared the performances of ChatGPT-4o, Claude 3 Opus, and Gemini Advanced across various question domains. The results of second-year periodontics residents served as a benchmark. Results: ChatGPT-4o, Gemini Advanced, and Claude 3 Opus significantly outperformed second-year periodontics residents across the United States, with accuracy rates of 92.7 percent, 81.6 percent, and 78.5 percent, respectively, compared to the residents’ 61.9 percent. The differences in performance among the AI models were statistically significant ( p < 0.001). Percentile rankings underscored the superior performance of the AI models, with ChatGPT-4o, Gemini Advanced, and Claude 3 Opus placing in the 99.95th, 98th, and 95th percentiles, respectively. Conclusion: ChatGPT-4o displayed superior performance compared to Claude 3 Opus and Gemini Advanced. The results highlight the potential of AI large language models (LLMs) as educational tools in periodontology and emphasize the need for ongoing evaluation and validation as these technologies evolve. Researchers should explore both the integration of AI language models into periodontal education and their impact on learning outcomes and clinical decision-making.Keywords
This publication has 10 references indexed in Scilit:
- Atypical gingival swellingThe Journal of the American Dental Association, 2023
- Performance of ChatGPT and GPT-4 on Neurosurgery Written Board ExaminationsNeurosurgery, 2023
- Journal of Dental Education Volume 87 Number 8/August 2023Journal of Dental Education, 2023
- The Role and Potential Contributions of the Artificial Intelligence Language Model ChatGPTAnnals of Biomedical Engineering, 2023
- Lessons learned from a novel 3-year longitudinal stepwise “Residents-as-Teachers” programJournal of the Chinese Medical Association, 2023
- ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid ConcernsHealthcare, 2023
- Artificial Intelligence in Dentistry: Chances and ChallengesJournal of Dental Research, 2020
- Recent Trends and Future Direction of Dental Research in the Digital EraInternational Journal of Environmental Research and Public Health, 2020
- High-performance medicine: the convergence of human and artificial intelligenceNature Medicine, 2019
- Could artificial intelligence make doctors obsolete?BMJ, 2018