Comparison of the Readability of Artificial Intelligence Applications in Periodontal Treatment

OpenAlex İngilizce

Objective: This study aims to compare the readability of responses provided by two prominent Large Language Models (LLMs), ChatGPT and Gemini, to frequently asked questions by patients regarding periodontal diseases. Materials and Methods: Ten common questions encountered by periodontology specialists at a university hospital were compiled. These questions were asked to both LLMs at 20 different times, and their responses were recorded. Readability was assessed using various indices, including the Automated Readability Index (ARI), Flesch Reading Ease, Gunning Fog Index, Flesch-Kincaid Grade Level, Coleman-Liau Index, SMOG Index, Linsear Write Formula, and FORCAST. Statistical analysis was performed using independent samples t-tests, with p < 0.05 considered significant. Results: Gemini produced texts with significantly higher reading levels—indicating greater complexity—compared to ChatGPT across several metrics: ARI (p=0.006), Flesch-Kincaid Grade Level (p=0.005), SMOG Index (p=0.003), and Linsear Write Grade Level (p=0.008). No statistically significant differences were found in Flesch Reading Ease (p=0.088), Gunning Fog Index (p=0.134), or Coleman-Liau Index (p=0.274). Conclusion: The findings indicate that Gemini’s responses are generally more complex and challenging to read than those generated by ChatGPT. This suggests significant differences in how these models adapt their outputs for a general audience. Relying on a comprehensive set of readability formulas is essential when evaluating the suitability of AI-generated dental health content for patient education.