The AAO have commented on Artificial Intelligence in Orthodontics.

Recent and rapid advances in artificial intelligence (AI) are likely to change clinical care in orthodontics.As with all technological advances, concerns arise regarding responsible use. The American Association of Orthodontists (AAO) recognized this issue and published a position paper this month in the AJODO. It is open access. Let’s take a look.

This is a guest post by Dr Klaus Batista, who is Adjunct Professor at the School of Dentistry Rio De Janeiro University, Brazil. He is developing a reputation on AI in orthodontics.

Introduction

Before starting, I’d like to clarify what a position paper means. One common definition is:

“A detailed report that recommends a course of action on a particular issue.”

It is relevant to point out that the recommendations in this position paper were developed by the AAO Task Force on Artificial Intelligence in Orthodontics. Although a position paper is not a high-level research paper, it is intended to provide relevant guidance for the responsible adoption of AI technologies in orthodontic clinical care.

Is this important?

AI has become a field of great interest worldwide. In orthodontics, there are several interesting uses of AI, such as orthodontic monitoring, cephalometric tracing, diagnosis, and treatment planning. Concerns arise about AI autonomy and the potential risk to patients from the lack of human participation in the process. Humans remain ultimately responsible for the final decision-making.

The Authors

The task force was composed of people from diverse backgrounds: private practitioners, academic orthodontics departments, the American Board of Orthodontists, AAO staff, and AI and digital companies. I will return to this later.

The Position Paper

In the introduction, the authors highlighted that the Task Force was formed because adaptive AI, a form of artificial intelligence, introduced a fundamental shift in clinical risk.

They grounded the position paper in three concepts: context of use (COU), the level of clinical risk involved, and the need for professional judgment by a licensed orthodontist.

They then emphasized that the central issue is how the presence of AI in orthodontics reshapes the orthodontist’s decision-making according to the function and degree of AI influence on patient care. To measure this influence, they divided the software into three types:

1. Clinical decision support software (CDS): provides information and recommendations without replacing professional judgement (cephalometric measurement);

2. Software as a medical device (SaMD): autonomously designs and delivers a device to be used clinically (bracket transfer tray);

3. Artificial intelligence/machine learning (AI/ML) software as a medical device in orthodontics: refers to SaMD that integrates ML capabilities (for example, a dental image analyzer used for remote treatment monitoring).

They then suggested six core principles to preserve professional authority, promote transparency in AI development, protect patient safety, and establish a principles-based framework:

Core Principle 1: AI governance and the human-in-command

Explains that AI governance is responsible for approving, validating, and monitoring AI systems to protect patient care. The human-in-command (HIC) concept puts the licensed orthodontist in charge of authorising and supervising AI use.

The call to action is that orthodontists must have the final word over all clinical decisions involving AI tools.

Core Principle 2: regulatory alignment and risk-based oversight

Explains that AI tools must follow regulatory requirements and require risk-based oversight according to the AI tool and its risk for harm.

The call to action asks that regulators, developers, and clinicians to jointly hold tool development and use to the highest international standards.

Core Principle 3: trustworthiness and transparency across the lifecycle

Explains that AI tools must be reliable, explainable, and consistent with transparency about how the models are trained, validated, updated, and monitored from the initial development through post-market use.

The call to action is directed to developers, who must build transparent, well-validated AI tools throughout their lifecycle; to regulatory bodies, to protect public trust; and to orthodontists, as the final decision-makers.

Core Principle 4: patient autonomy

This principle talks about how patients should be told by the orthodontist when AI tools are used in clinical decision-making. Patients should authorise the use of identifiable protected health information for AI development, validation, or retraining, when required.

The call to action is directed to clinicians to document AI use in patient records to clearly communicate so patients receive explanations of how AI supports their care and how their data may be used. State dental boards should also align their policies with national and international AI governance frameworks to protect patient rights and data privacy. Training programs should give clinicians the skills to use AI responsibly, explain it to patients in plain language, and keep data secure.

Core Principle 5: Education and Clinical Artificial Intelligence Competency

This principle recommends that orthodontists must receive education in AI throughout the profession to ensure AI strengthens professional judgement.

The call to action is to add AI coursework to dental school and residency training, provide continuing education courses (AAO), stimulate vendor-sponsored training programs, and help faculty build up their own AI knowledge so they can teach it.

Core Principle 6: Operational Integration and data privacy

This principle explains that safely working AI into a practice takes ongoing attention to safety, interoperability between systems, and protecting patient data at every stage.

The call to action is to use standardised data-exchange protocols, require compliance with HIPAA, the HITECH Act, and ISO/IEC security standards, vendors should be transparent about how they collect and use patient data, smaller practices need funding or technical help to keep up, and clinicians and staff need ongoing training in AI data ethics and cybersecurity.

Finally, they lay out a three-phase approach for evaluating and implementing AI tools: evaluate, implement, monitor, arguing that orthodontists should adopt new AI tools deliberately rather than ad hoc.

The authors highlighted some limitations, as the position paper does not establish legal responsibilities and only reflects the AAO’s current position.

They concluded that: 

• “The framework and core principles outlined in this document provide a roadmap for the safe, equitable, and effective adoption of AI in clinical orthodontics.”
• “Orthodontists remain accountable for technology selection and for the clinical use and oversight of AI tools at the point of care, including review of available validation evidence.”
• “Developers share responsibility for the ethical design, transparency, and secure deployment of AI tools.”
• “Data integrity and patient privacy must not be compromised.”
• “Continuous education is essential to sustain clinical readiness in a rapidly evolving digital landscape.”

The last paragraph started with a thoughtful and forward-looking phrase:

“The AAO envisions a future in which AI integration strengthens both the science and the art of orthodontics, advancing innovation while preserving humanity in care”.

What did I think?

As we have limited space, I will focus on some relevant aspects that I observed.

In my view, this position paper represents a useful guide for those interested in using AI responsibly. Still, I’d like to start with the task force composition. It was clear that they tried to form a diverse panel. However, three authors came from AI and digital companies (CADflow.ai, Dental Monitoring, and Align Technology) with a direct commercial interest. Even though all the authors declared no conflict-of-interest, I’d still flag it. The readers deserve to know.

Additionally, it would be nice if lay people had participated in the panel.

Going back to the position paper content, I considered that one of the most relevant aspects is the human-in-command approach, which places professional judgement at the centre of AI integration.

The position paper also made me think about a scale of AI use, ranging from limited AI use with substantial human participation in the decision-making process to extensive AI use with limited human participation. This is referred to as the AI context of use (COU), and it can help define the risks for patients receiving treatment: for now, the less human participation there is, the greater the potential risk to the patient, though that balance could shift as AI becomes more assertive. Thus, patients should be informed about the extent to which AI is being used in their care.

Another relevant contribution of this position paper is helping to understand AI adoption as a shared responsibility among clinicians, developers, regulators, and educators.

Lastly, I strongly agree that AI literacy must be included in the education of new professionals and in faculty development to ensure that patients are treated ethically and can benefit from advances in AI.

Final thoughts

Although this position paper is extremely relevant, it is also extensive, and I am concerned that this information may not reach the orthodontic community, developers, regulators, and educators worldwide.

This is because some regulations are specific to the USA and may not apply to other countries. Nevertheless, I believe that most of the content presented can be useful and applicable in other nations. It would be valuable if other associations developed position papers with a broader, worldwide perspective on AI.

For orthodontists, the key message is clear: understanding AI is no longer optional, but its use must remain transparent, accountable, and focused on patient safety.

With the presence of AI, it is likely that we are entering a new era called augmented orthodontics, in which orthodontists can offer patients a better healthcare experience.

Disclosure: I used AI tools to help verify this summary against the original paper and to polish the writing. The analysis and opinions are my own.