Related Experiment Video
Updated: Feb 15, 2026

Generation of Transgenic C. elegans by Biolistic Transformation
Published on: August 23, 2010
Performance and repeatability of different models of Chat Generative Pretrained Transformer (GPT) in generating
Nawal M Alharbi1, Reham B Osman2
1Associate Professor at Department of Prosthetic Dental Sciences College of Dentistry, King Saud University, Riyadh, Saudi Arabia.
Objectives:
To evaluate the performance and repeatability of different artificial intelligence chatbots (ChatGPT-4o, ChatGPT-o3 and ChatGPT-5) in generating the most optimal design for a removable partial denture framework for a given clinical case.
Methods:
Nine gypsum models of partially edentulous clinical cases were randomly selected. All models were digitized, and the digital files were imported into an AI-chatbot (ChatGPT-4o, ChatGPT-o3 and ChatGPT-5) to generate a removable partial denture framework design using one search prompt at two time points (5 days interval in-between). The generated images were evaluated following eleven binary evaluation criteria. Two experienced prosthodontists evaluated all generated images and scored each criterion. Overall performance of AI chatbots were calculated as accuracy percentage. Cochran's Q test with post-hoc McNemar's tests were used to compare the responses between the tested ChatGPT models. Repeatability was assessed using Cohen's κ and percentage agreement, and agreement between the two evaluators (α=0.05).
Results:
Cochran's Q test showed significant differences between different chatbots (Q = 12.333, P = 0.002), McNemar's test showed that ChatGPT-o3 significantly outperformed ChatGPT-4o and ChatGPT-5, and ChatGPT-4o outperformed ChatGPT-5. Repeatability analysis showed high agreement in ChatGPT-4o, ChatGPT5, whereas lower agreement was observed with ChatGPT-o3.
Conclusions:
Within the limitation of the study, the tested models showed low performance in generating clinically acceptable RPD designs and were unable to provide repeated results when rechecked. Chatbot GPT-4o, ChatGPT-o3 and the more advanced ChatGPT-5 exhibit very limited ability to generate clinically acceptable RPD designs.
Clinical Significance:
The use of AI chatbots to design RPD frameworks is not recommended due to the complex nature of prosthetic planning presenting various biomechanical and biological considerations.
More Related Videos
09:44Generation of a Rat Model of Acute Liver Failure by Combining 70% Partial Hepatectomy and Acetaminophen
Published on: November 27, 2019
08:39Generation of Electronic Cigarette Aerosol by a Third-Generation Machine-Vaping Device: Application to Toxicological Studies
Published on: August 25, 2018
Related Concept Videos
Bioequivalence Experimental Study Designs: Repeated Measures, Cross-Over, Carry-Over, and Latin Square Designs
Inverse z-Transform by Partial Fraction Expansion
To begin the process, the poles of the function are identified and the function is...
Group Design
DC Generator
Next-generation Sequencing
Next-Generation Sequencing Methods
Although all next-generation methods use different technologies, they all share a set of standard features....
Generation Time