I was very interested to come across a headline recently that suggested ChatGPT had diagnosed a rare condition and saved a woman’s life. It adds fuel to the already widespread conversation about the role of artificial intelligence (AI) in modern healthcare. As dental professionals, we have broadly accepted that AI is here to stay and we accept the benefits it affords. However, we remain aware of its limitations. Has the time come to discuss AI in more detail with our patients in order to educate and support them in its use when it comes to improving health?
The story in question
The news story that provoked this chain of thought was about a young woman who struggled with her health for many years.[i] Her medical history combined with new symptoms led to misdiagnoses by doctors, who were struggling to determine and treat her condition. In putting her symptoms into ChatGPT, it produced a list of potential causes, one of which a doctor thought plausible and confirmed the diagnosis upon genetic testing.
The implications of this being in the general media are multiple. Firstly, it does highlight the positive impact that AI can have in supporting healthcare provision. Secondly, it shows how useful a resource AI-assisted platforms can be for individuals who are conducting research. They are highly accessible tools for people who want to engage more with their health and perhaps take some control when it comes to modifiable risk factors for common conditions.
A risky business
However, this is only one side of the story.
Though it is developing at a rapid pace, current AI capabilities fall short of a consistent and accurate medical diagnosis. A study led by the Oxford Internet Institute and Oxford University found that use of Large Language Models (LLM) gave individuals no more or better information than a traditional online search – or their own judgement.[ii] It showed that most people didn’t know what to tell the AI model in order to get the advice they wanted. It also demonstrated the variety in answers that LLMs can give to the same situation, based on slightly different presentations of the facts. When a combination of good and bad guidance was issued by the AI platform, users struggled to determine which was which.
The issue here is that a growing number of people in the UK are treating AI-driven tools like a digital doctor. Almost one in 10 are now using AI-powered chatbots for health advice.[iii] Approximately 37% of UK adults are turning to the same platforms for mental health and wellbeing support.[iv] As such, a considerable proportion of the population could be exposed to inaccurate and unreliable medical advice at any given time.
Blurred lines
Despite the risks, it is unrealistic to expect the disuse of AI, especially given the current economic climate and widespread waiting times for medical care. Instead, I am of the opinion that we need to educate patients on how to use it safely and effectively. After all, dental professionals use AI-assisted technology in the practice every day. It would be hypocritical of us to promote our cutting-edge AI-supported facilities, only to tell patients that AI is dangerous.
Instead, they should adopt a similar approach to us. AI is a second opinion, a verification, a tool for idea generation. It should not be relied upon solely, without personal judgement, additional fact checking, or professional insight. With a hybrid approach, AI-driven platforms could be used more effectively to accelerate diagnostics and put patients’ minds at ease.
A second opinion for the second opinion?
Putting AI to the test for myself, it seems to provide fairly relevant advice if asked what to do with persistent pain following endodontic treatment. It recommends returning to the clinician for further assessment and even suggests retreatment as a potential outcome.
However, clinical diagnosis will require much more than written (or verbal) information, potentially necessitating further radiographic investigation. Even then, retreatment can be complex and requires specialist skills to deliver. As ChatGPT is inadequate in such a situation, the specialist team at EndoCare would be more than happy to help. We have several decades of genuine experience in the field, offering a broad range of advanced endodontic procedures from microsurgery to retreatment. Our referral process is quick and simple – just complete the form on the website to get started.
Embracing opportunity – safely
AI is not something to be feared, but it is not yet ready to be trusted explicitly without outside influence. When used safely, modern platforms provide a helpful resource for people to educate themselves about the risks and symptoms of health conditions. The key to preventing inaccuracies and misinformation from causing any harm is understanding that recommendations should always be confirmed by the relevant professional.
For further information about the endodontic referral services available from EndoCare, please call 020 7224 0999 or visit the NEW website www.endocare.co.uk
Author: Michael Sultan
[i] Herbert D. ChatGPT ‘uncovered woman’s rare condition’ after years of misdiagnosis. April 2026. https://www.bbc.co.uk/news/articles/cx24njzzkgjo [Accessed July 2026]
[ii] Bean, A.M., Payne, R.E., Parsons, G. et al. Reliability of LLMs as medical assistants for the general public: a randomized preregistered study. Nat Med 32, 609–615 (2026). https://doi.org/10.1038/s41591-025-04074-y
[iii] Non-technical summary: section 2 – information on how AI is used in health and the current regulatory frameworks. Medicines & Healthcare products Regulatory Agency. June 2026. https://www.gov.uk/government/calls-for-evidence/regulation-of-ai-in-healthcare/non-technical-summary-section-2-information-on-how-ai-is-used-in-health-and-the-current-regulatory-frameworks [Accessed July 2026]
[iv] Over one in three using AI Chatbots for mental health support, as charity calls for urgent safeguards. November 025. https://mentalhealth-uk.org/news-and-insights/over-one-in-three-using-ai-chatbots-for-mental-health-support-as-charity-calls-for-urgent-safeguards/ [Accessed July 2026]


