,

The Future of Dental Care: Ethical AI Guidelines

The potential benefits of incorporating AI into dental practice are clear. Improved workflows, supported clinical decision making and treatment planning are all possible quality benefits. Dental practice is already a litigious area and despite the benefits, there are risks. The use of AI is another area that can be incorporated into legal claims. Dental Protection,…

robot hand reaching out to human hand

The potential benefits of incorporating AI into dental practice are clear. Improved workflows, supported clinical decision making and treatment planning are all possible quality benefits.

Dental practice is already a litigious area and despite the benefits, there are risks. The use of AI is another area that can be incorporated into legal claims. Dental Protection, who have dental members in many countries, have produced guidance on using AI in practice.

The guidance is in two parts. The first part, ‘Informed’, deals with ethical decision making. The second part, Records’, focuses on recording the use of AI.

  • Informed: Using Informed as an acronym, the guidance works its way through deciding where to use AI; understanding its role and what it can and cannot do; reviewing AI output and deciding how and if to incorporate it into treatment plans; communicating with patients, and ensuring that the dentist’s duty of care is being met.
  • Records: This section refers to the way that the use of AI is recorded. This includes when and how it is used; comparison of the AI output to the patient presentation and investigations; documenting AI’s role, and ensuring that the record documents that patient is aware of how and where an AI has contributed to their treatment so that they were able to make an informed choice.

This a very useful guide that covers a lot of territory. Many of the recommendations will feel comfortable to any dentist. Comparing what an AI says to what you expected, based on the presentation, history and investigations, makes obvious sense. Recording when and how an AI is used, also seems like best practice.

The challenges lie in some of the detail. These include:

Considering any potential bias in how the AI was trained: This is an important issue because using datasets with a limited demographic can affect the applicability to other populations, This is a challenging area for individual practitioners. Outside of published models developed by academics, practice teams will be reliant on system developers and software suppliers to provide an account of which datasets were used to develop the AI. There are efforts to improve the diversity of dental datasets, but in the meantime, the Dental Protection guidance suggests that dental teams should ask questions about the datasets used and their applicability to their population, when incorporating a new AI application.

Understanding how the AI arrived at its recommendation: The Dental Protection guidance states that the team should:

Be able to understand, interpret and communicate how the AI system arrived at its output.

How demanding a requirement this is, depends on what it means. In broad terms, every dentist can learn to explain how an AI is taught, and how it then uses this learning to produce recommendations. Understanding the detail of how an individual AI was taught is a much higher bar. Knowing how it makes a specific patient recommendation based on predictive analytics is impractical.

Communicating known risks: This overlaps with the points above and is relatively manageable. System providers should be able to give teams information on the accuracy of their systems. This still needs interpretation. Research on one commercial charting system found that agreement with experienced clinicians was good, but varied by tooth location and patient age and gender. Scores for full-mouth reports using the same system were lower. Awareness of the accuracy of systems being used in a practice will help clinicians to interpret AI output and to guide their patients accordingly.

The Dental Protection guide is very useful, and fills a existing gap in practical guidance, particularly in recording how an AI was used in clinical treatment. The guidance is accessible and the principles could be put into practice in any country or service setting.

Reference:

Dental Protection (2025) Artificial Intelligence: Your dentolegal guide to using AI in practice. The Safer Practice Framework. Leeds, Dental Protection. Available at: AI Safer in Practice

Photo by cottonbro studio on Pexels.com

Leave a comment