Dr. Teck has been passionate about prevention, hygiene, and diet since dental school. Over the years as his practice grew, he and his office team explored and tried many caries detection and prevention methods—chemical tests of saliva and blood; imaging devices; patient questionnaires about habits, diet, and medications; careful examinations of dental anatomy, tooth positions, and previous restorations; etc. Eventually, his customary approach came down to Caries Management by Risk Assessment (CAMBRA). However, even that was limited in its ability to detect, much less prevent, caries. Many patients did not keep appointments. Children he screened in community schools did not have adequate income or third-party payments to address all their oral needs. And many patients simply got lost due to the situations of complex family dynamics, moving from the area, and so on.
Last week, Dr. Teck came back from a seminar that talked about the revolutionary advantages of Artificial Intelligence (AI). It was already evident in automobiles’ computer-initiated safety maneuvers and even driverless cars, in computer face recognition, and in a growing number of medical data analyses that claim to be aiding doctors in making better decisions. In dentistry, the seminar identified AI technologies that were already being used to aid diagnostics, mainly in dental imagery, but also in collecting and managing complex clinical datasets. More AI-based systems would be coming to the dental market soon that would claim to be better than the “average dentist” in detecting diseases and making predictions as to their courses and the best treatment choices. Despite what he was told, Dr. Teck was not certain if the underlying evidence for these systems would be as strong as it should be or if these would be just more costly tools that would be no better than his already generally quick clinical sense of each situation at hand. He questioned if AI research in dentistry could even meet the level of the best evidence-based medicine when, and/or if, it ever faced the rigors of standardization. So, he wondered if he was just being caught up, yet again, in the hype and promises of the new and unproven. However, he hoped that getting involved with perfecting AI diagnostics might make his care for his patients even better and might also make the development of AI-aided dental treatments more likely.
No one at the seminar asked a question about ethics. But Dr. Teck also wondered if introducing AI technologies into dental offices would raise any ethical questions.
At present, the AI tools that are currently available or in development for dental use are focused on assisting dentists’ diagnoses of their patients’ needs. If AI tools to assist or replace dentists’ therapeutic interventions begin to be developed, then ethical questions about their proper use will need to be considered when that happens. This article will focus on the ethical issues that concern dentists’ use of AI tools in dental diagnosis.
Four kinds of ethical issues are relevant to dentists’ use of AI diagnostic tools. The first issue is the most obvious: namely, whether the AI tool itself actually provides better or at least equally dependable diagnostic data to the dentist and whether the dentist properly uses the AI diagnostic tool itself so that the information it provides is correct within the statistical limits of the tool. In order to use an AI diagnostic tool in an ethically proper manner, the dentist needs to understand enough about how the tool achieves its results to be assured that its results are dependable and needs to use it in such a way as to routinely obtain dependable output from it. Naively believing the hype and the marketing materials and then merely following the instructions in the instruction booklet will not enable the dentist to be assured that the tool’s output is dependable. A dentist who fails to meet these ethical standards obviously fails in his or her professional duty to provide expert professional oral health care to the patient.
A second ethical issue concerns the possibility that the dentist might rely so heavily on the AI tool’s statistically based results that his or her professional diagnostic judgment about a unique patient remains inactive. The dentist should not make a therapeutic recommendation solely from what the AI output points to but rather make a personal-professional recommendation about the therapy that is best for this unique patient. The dentist can take into account the AI-developed statistically based data along with the dentist’s own other diagnostic efforts about this patient’s particular needs. This is because the output of an AI diagnostic tool is necessarily statistically grounded and, therefore, does not necessarily connect with this unique patient unless the dentist employs his or her professional expertise to make that connection properly. Again, a dentist who fails to meet this ethical standard fails in his or her professional duty to provide expert professional oral health care to the patient.
The third ethical issue concerns how the dentist communicates with the patient about the diagnostic data provided by the AI tool. There are 2 aspects to this ethical concern. The first part of the process is providing the patient with adequate information so that the patient can make an autonomous, reasoned, well-informed decision about treatment as the information provided by the AI tool is necessarily statistical. Therefore, it should be made clear to the patient when the dentist shares the data provided by the AI tool that, while this is a useful diagnostic tool, its output is at best statistical, and, therefore, it does not automatically determine what treatment would be best for this patient at this time. The patient still needs the dentist to interpret for him or her to what extent the AI tool’s output applies to his or her condition and then to work with him or her as they determine (ideally working together) which of the treatments within the standard of care (or no treatment at all) ought to be chosen.
In addition, if a dentist were to give excessive weight to the output of an AI diagnostic tool, the dentist could easily communicate to the patient that the AI tool is a better diagnostician than the dentist or, more generally, that replacing dentists with appropriate AI tools would lead to better oral health care for patients. Dentistry’s contribution to society as a profession is based on dentistry’s special expertise, not just general and statistical knowledge that most people don’t have (although in today’s world, they can get some of it online), but especially the expertise in applying that body of knowledge to protect and improve the oral health of unique human beings. AI diagnostic tools may be able to add to that body of knowledge, but they cannot determine what is the best course of treatment for a unique human being. For this reason, a dentist would be failing in his or her obligations to the profession and to the larger society if he or she handed over or appeared to hand over the provision of diagnoses completely to an AI diagnostic tool. Doing so would be sending a false message about oral health and dental expertise to that dentist’s patients.1
A fourth ethical issue concerns the dentist’s obligation to provide the patient with appropriate information about the procedures used in the patient’s care whenever the dentist is still learning how to use a new piece of technology. Patients who seek care from a dentist can reasonably be assumed to give implied consent to any diagnostic procedures the dentist uses to assess the condition of their oral cavities, but only because they assume that the dentist is both thoroughly trained and adequately experienced in whatever diagnostic procedures he or she employs. But this implied consent does not automatically cover procedures, including diagnostic procedures, that the dentist is just learning to use in practice.
It would ordinarily be ethically sufficient for the dentist to say something like “I am using a new diagnostic tool today to determine [whatever the AI tool is for]. I am just learning to use it, and I will be double checking what it tells me against what I have determined from my own examination of your [whatever]. It is supposed to give me even more precise measurements of [whatever], and that would make my diagnosis even more dependable. But I wanted you to know about it since it’s not something you’ve seen me use before.” This gives patients an opportunity to ask a question if they wish and even to decline having the dentist use it (though this is likely to be rare). But much more importantly, it provides patients with information they have a right to know and that the dentist has an obligation to provide; namely that, at this point, the dentist’s use of this AI tool is not yet based on thorough education and adequate experience.2
New technologies are never ethically neutral from a professional perspective, so it is important to follow Dr. Teck’s lead in our case. It is time to ask ourselves in what ways a dentist’s professional obligations are affected by each new technology and procedure that becomes available.
1. Ozar DT, Sokol DJ, Patthoff DE. The wrong message and why it matters. In: Ozar DT, Sokol DJ, Patthoff DE. Dental Ethics at Chairside: Professional Obligations and Practical Applications. 3rd ed. Washington, DC: Georgetown University Press; 2018:279-302.
2. Ozar DT, Sokol DJ, Patthoff DE. New technologies and procedures. In: Ozar DT, Sokol DJ, Patthoff DE. Dental Ethics at Chairside: Professional Obligations and Practical Applications. 3rd ed. Washington, DC: Georgetown University Press; 2018:292-295.
Dr. Patthoff is a general dentist in Martinsburg, W Va. He was a principal research investigator at the Martinsburg VA dental clinic and is a past president of the West Virginia Dental Association and the American Society for Dental Ethics (ASDE) (formerly the Professional Ethics in Dentistry Network). He co-authors a dental ethics column in AGD Impact magazine and is an ethics consultant to the ADA’s Council of Ethics Bylaws and Judicial Affairs. He is also a co-author of Dental Ethics at Chairside and is editor-in-chief of the Journal of Laser Dentistry. Dr. Patthoff chaired the dental sessions of Engineering Conference International’s 3 conferences on light-activated tissue regeneration as well as 2 sessions on photobiomodulation for the Optical Society of America. He also chairs the ethics committees of the Academy of Laser Dentistry, WVU Medicine’s Berkeley Medical Center, and the George Washington Institute of Living Ethics. Dr. Patthoff is also a liaison to the International Dental Ethics and Law Society from the ASDE. He can be reached at email@example.com.
Prof. Ozar is professor emeritus of philosophy at Loyola University Chicago, where he taught professional and healthcare ethics from 1972 to 2015. He developed and taught a course in professional ethics in dentistry at Loyola’s School of Dentistry from 1981 until the school closed in 1994 and has lectured and consulted for dozens of dental schools and dental professional societies. He founded the American Society for Dental Ethics (ASDE) in 1987 and was its executive director until 2005. He was the founding editor of “Issues in Dental Ethics,” a regular feature of the Journal of the American College of Dentists, and has authored numerous additional articles on dental ethics, professional ethics, ethics education, and contemporary social issues, including more than 30 essays with co-author Dr. Patthoff in AGD Impact. The third edition of Dental Ethics at Chairside was published by Georgetown University Press in spring 2018. He is an Honorary Fellow of the American College of Dentists and the American College of Legal Medicine, and, in his honor, along with the late Thomas Hasegawa, DDS, the ASDE named its annual student dental ethics prize “The Ozar/Hasegawa Ethics Award.” He can be reached at firstname.lastname@example.org.
Disclosure: The authors report no disclosures.