Abstract
Objective
Generative Artificial Intelligence (GenAI) is revolutionizing medical and dental education. However, most off-the-shelf chatbots lack the necessary accuracy and comprehensiveness required for clinical teaching, particularly in advanced specialties such as implant dentistry. Despite technological advancements, students’ knowledge from traditional learning resources is essential for guiding the AI systems in cultivating comprehensive and ethical clinical decision-making. To address this issue, we aimed to develop a customized chatbot, meticulously trained with verified teaching materials, and to evaluate its effectiveness as an AI tutor in enhancing the learning experience of dental students.
Method
Dental students were randomly assigned to two groups to study the same clinical case within a problem-based learning (PBL) setting. Both groups engaged in similar learning processes, with the exception that only one group utilized our custom-trained chatbot as their AI tutor during the PBL. The knowledge acquisition and understanding of the students were evaluated by comparing their scores on standardized questions related to implant dentistry. To investigate the influence of prior clinical experience on learning outcomes, students with varying levels of clinical background were invited to participate in the study.
Results
Through the integration of reliable and credible resources, followed by multiple rounds of refinement, the AI tutor was developed to provide detailed and accurate responses to technical questions in both text and voice modes. We found that students with access to our custom-trained AI tutor achieved significantly higher knowledge scores. Moreover, students without the AI tutor were found to be less likely to pose questions pertinent to the learning objectives of the PBL case. Additionally, senior students were found to benefit more from the use of the AI tutor, probably due to their more extensive prior knowledge and clinical experience. Collectively, our data suggested that the application of a custom-trained AI tutor into PBL could significantly contribute to enhanced learning outcomes, especially among senior dental students.
Conclusion
This study demonstrates the potential of GenAI, after being properly trained with verified teaching materials, to serve as an effective tutor in facilitating PBL. We anticipate that the custom-trained AI tutor can be adopted in more teaching settings to improve learning outcomes and to relieve the teaching burden.
Presenter(s) / Author(s)
