AI model performance on dental examinations
Investigators have examined how generative artificial intelligence tools fared when taking various dental exams.
In a study published in Imaging Science in Dentistry, the investigators analyzed the performance of the ChatGPT 3.5 and 4.0 AI models on 253 questions from U.S. dental examinations, including the Integrated National Board Dental Examination, Dental Admission Test and Advanced Dental Admission Test. The results were evaluated against the official answer sheets.
The investigators found that the accuracy scores for both ChatGPT 3.5 and 4.0 were comparable for the first two tests. For instance, the models answered 80% of knowledge-based questions and 66% to 69% of case history questions accurately on the Integrated National Board Dental Examination as well as 66% to 83% of knowledge-based questions and 76% of case history questions on the Advanced Dental Admission Test. However, on the Dental Admission Test, ChatGPT 4.0 was more accurate compared with ChatGPT 3.5, scoring 94% vs 83% on knowledge-based questions, 57% vs 31% on mathematical analysis items and 100% vs 82% on comprehension questions.
The performance of the two models, particularly ChatGPT 4.0’s performance on comprehension questions indicated that the models could be highly trainable on certain topics. Nonetheless, both models may require improvements in mathematical analysis.
Read more: Imaging Science in Dentistry
The article presented here is intended to inform you about the broader media perspective on dentistry, regardless of its alignment with the ADA's stance. It is important to note that publication of an article does not imply the ADA's endorsement, agreement, or promotion of its content.