AI chatbots align with implant prosthodontic evidence

Dental Tribune International

Tue. 30. June 2026

save

PIRACICABA, Brazil: In implant prosthodontics, the growing body of research that clinicians must navigate raises the possibility of using large language model-based artificial intelligence (AI) tools to help organise this information and support clinical reasoning, but their alignment with systematically appraised evidence remains unclear. A new study has examined this issue by comparing responses from ChatGPT and Google Gemini against conclusions from recent systematic reviews and found a high level of agreement, but also differences in the certainty levels reported for the platforms’ answers.

The researchers compared responses generated by ChatGPT and Google Gemini with the conclusions of 74 systematic reviews published between 2023 and 2025. The reviews covered five domains: prosthesis design, implant components, biological and mechanical complications, clinical protocols and prosthetic retention.

To create a benchmark, the authors formulated clinical questions from the objectives of the systematic reviews and used the conclusions of the reviews as expected answers. They then submitted the questions to both AI platforms using identical prompts. The responses were then independently evaluated against the review conclusions by two blinded assessors.

Across all five clinical domains, ChatGPT and Gemini generally generated answers that aligned closely with the conclusions of the systematic reviews, and there were no statistically significant differences in agreement between the platforms. Previous research has similarly reported that some AI chatbots can provide accurate and consistent responses to common prosthodontic questions, although performance varied depending on the platform and the complexity of the topic.

According to the researchers, the findings point to a potential role for AI tools to help practitioners navigate the expanding body of implant dentistry research more efficiently. However, important differences emerged in the certainty levels the platforms gave for their answers. ChatGPT most frequently described its answers with moderate confidence, whereas Gemini tended to express high confidence. While this did not affect agreement with the review conclusions, the authors cautioned that confidence levels should not be interpreted as a measure of evidence quality.

The study also highlighted broader concerns surrounding the use of AI in healthcare. Large language models can generate plausible but inaccurate responses, and users may place undue trust in highly confident answers. The authors concluded that AI chatbots show promise as support tools for evidence synthesis in implant prosthodontics but that their integration into evidence-based clinical workflows requires careful supervision and critical interpretation in clinical settings.

The article, titled “Alignment of artificial intelligence-generated responses with systematic reviews in implant prosthodontics”, was published online on 2 June 2026 in the Journal of Prosthetic Dentistry, ahead of inclusion in an issue.