Artificial Intelligence In Dental Exams: A Performance Comparison Of ChatGPT, DeepSeek, And Students
TL;DR
This study aims to comparatively assess the performance levels of ChatGPT and DeepSeek in the multiple-choice Dentomaxillofacial Radiology examination administered to third-grade students, both internally and in relation to one another.
Abstract
Aim
The utilization of Artificial Intelligence (AI) systems adept at searching web data to address diverse inquiries is progressively escalating. This study aims to comparatively assess the performance levels of ChatGPT and DeepSeek, both of which are gaining prominence in academic circles, in the multiple-choice Dentomaxillofacial Radiology examination administered to third-grade students, both internally and in relation to one another. The results will also be compared to the performance levels of third-year dentistry students. MATERIALS and
Methods
The 25 multiple-choice questions from the Dentomaxillofacial Radiology examination given to third-year students at Gülhane Faculty of Dentistry were submitted to the ChatGPT 4o and DeepSeek programs at different intervals. The outcomes of these three groups were assessed comparatively, and the capabilities and constraints of AI in medical examinations were examined.
Results
The findings indicate that the consistency of the ChatGPT and DeepSeek algorithms is statistically significant (p<0.001). Despite both programs attaining a passing score in this examination, their performance, in comparison to the students, was akin to that of an ordinary student.
Conclusion
This study may enhance research on the efficacy of AI algorithms like ChatGPT 4o and DeepSeek in dentistry. Nonetheless, it is essential to acknowledge that the practice of medicine is not exclusively governed by rigid regulations; it also necessitates critical thought and interpersonal engagement, highlighting its creative aspect as a dynamic profession.