
AI in Dentistry: Ensuring Fairness in Patient Care
AI systems are predominantly trained on data that is created by humans; although recently concern is growing that AI is now using other AI generated content to develop.
What is clear is that this data, and those who input it, contain inherent biases. This can lead to algorithms acknowledging and replicating those same biases for future recommendations.
The Problem of Bias in AI
Currently training datasets are overwhelmingly in English, and represent Western cultures. This is being replicated in AI.
For example:
- Generation models have been found to excessively replicate gender stereotypes
- Occupations such as pilots are often shown as male
- Hairdressers are often shown as female
- White individuals are overrepresented across most domains
Exceptions tend to appear only in limited roles such as pastor or rapper.
AI Within the Dental Profession
In this edition of our AI Within the Dental Profession series, we consider the real risks of discrimination and bias within AI.
The good news is that there is a widespread recognition that data bias is a real and present problem, with databases themselves discriminating in the information they provide.
The real question is: what is being done to protect us from AI discrimination?
Legal and Ethical Responsibilities
Health services must be delivered fairly, without discriminating against patients on the grounds of a protected characteristic such as:
- Race
- Age
- Gender
- Sexual orientation
However, algorithms within the health sector have historically been found to contain racial bias, underestimating the needs of patients from marginalised racial and ethnic backgrounds.
Real-World Evidence of AI Bias
We know this as the US healthcare system uses commercial algorithms to guide health decisions. Previous studies had found evidence within a widely used algorithm that black patients, who were assigned the same level of risk as white patients, were far sicker.
This stemmed from the algorithm using health costs as a proxy for health needs. Less money is spent on black patients who have the same level of need, thus, the algorithm falsely concluded that black patients are healthier than the equally sick white patients.
The Office of the United Nations High Commissioner for Human Rights (OHCHR) noted in a report dated 30 July 2024 that:
“bias from the past leads to bias in the future”
This can result in discriminatory outputs based on:
- Race
- Gender
- Age
- Disability
With harmful effects across sectors including healthcare, education, and finance.
AI Discrimination Beyond Healthcare
Discrimination within AI is not a new phenomenon.
- In June 2020, Microsoft barred facial recognition sales to US police due to fears of racial bias
- UK police continue to use facial recognition technologies
- South Wales Police were found to have unlawfully used Automatic Facial Recognition in 2020 due to failure to assess bias
More recently:
- In March 2024, a black Uber Eats driver received a pay-out
- Facial recognition checks prevented him from accessing work
- The Equality and Human Rights Commission raised concerns about loss of income due to AI bias
How Can Dental Practices Mitigate Risk?
There are bias mitigation techniques which exist already, however, they are not yet reliable. Robust mitigation techniques remain in the early stages of development and are unable to keep up with AI’s rapid continued growth.
Key Steps for Practices
Practices should:
- Document their approach to bias and discrimination mitigation from the outset and throughout
- Ensure senior management signs off on chosen approaches
If considering replacing traditional decision-making systems with AI entirely:
- Run both systems concurrently for a period of time
- Audit differences in decisions affecting protected groups
Ongoing Responsibilities
- Invest time and resources into continuous staff training
- Ensure teams understand risks and can identify issues early
- Foster a proactive working environment
Ultimately, and finally, a diverse workforce is a powerful tool in recognising and managing bias and discrimination, not only within AI systems, but the practice more widely.
Need Advice on AI and Compliance?
If you need clear, practical advice on how to manage AI within your business, don’t hesitate to get in touch. A member of our trusted team will be able to advise on your legal position. Contact JFH Law today.
Please note that the information contained in this article was correct at the time of writing. There may have been updates to the law since the article was written which may affect the information and advice given therein.
[1] Racism and AI: “Bias from the past leads to bias in the future” | OHCHR
[2] Microsoft bars facial recognition sales to police – BBC News






