Exploring the potential of artificial intelligence chatbots in prosthodontics education

Area of Science:

• Dental Education
• Artificial Intelligence in Dentistry
• Prosthodontics

Background:

• Artificial intelligence (AI) chatbots are increasingly prevalent.
• Their performance in specialized dental fields like prosthodontics is largely unevaluated.
• This study assesses AI chatbot accuracy on prosthodontics exam questions.

Purpose of the Study:

• To evaluate the performance of widely used AI chatbots.
• To assess their accuracy in answering prosthodontics questions from the Dentistry Specialization Residency Examination (DSRE).

Main Methods:

• 126 DSRE prosthodontics questions across seven subtopics were used.
• Five AI chatbots (ChatGPT-3.5, Gemini Advanced, Claude Pro, Microsoft Copilot, Perplexity) were tested.
• Statistical analyses (chi-square, z-tests) compared accuracy rates.

Main Results:

• Microsoft Copilot demonstrated the highest overall accuracy at 73%.
• Copilot significantly outperformed Perplexity (54.8% accuracy).
• Accuracy varied by subtopic, with dental implantology (75%) highest and removable partial dentures (50.8%) lowest.

Conclusions:

• Microsoft Copilot exhibited the best performance among the tested AI chatbots.
• AI chatbots show potential as supplementary educational tools in prosthodontics.
• Current AI models have limitations for comprehensive and reliable use in dental education, necessitating future advancements.