I first wrote about artificial intelligence (AI) technology in medicine in my October 2023 column. Since then, restructured hospitals in Singapore have been leading the way in the use of AI for clinical documentation and in certain aspects of diagnosis and outcome prediction. Similarly, in the private sector, clinic management systems have begun to incorporate AI scribes into their offerings to aid in documenting clinic consultations. I have not embarked on the use of an AI scribe yet, as I am still wondering how to tell my patients that their voices will be recorded and stored. Currently, I use AI mainly for research – I sometimes use ChatGPT to help in searching for the medical information that I require for managing my patients.
In 2025, I attended a SkillsFuture course on utilising generative AI for improving productivity. Among other points, I learnt about the principle of prompt engineering, which is the process of designing text inputs to guide the AI model to produce optimal responses or answers. I also realised that there were other generative AI platforms that could also produce images, music and even slide presentations, and I learnt about the limitations of AI in terms of potential biases and their tendency to provide incorrect information (commonly referred to as AI hallucination). Lastly, I was reminded not to upload private or confidential information when using AI.
I believe that just as the Internet provided the means for us to obtain information at our fingertips, AI can enable us to be more efficient in our work. Even if we prefer to do things the old-fashioned way (which I am guilty of, as I am writing this article without the use of AI), we should at least be aware of what AI can and cannot do, like what I learnt at the SkillsFuture course.
To illustrate the point, I recently had a patient who consulted me, looking for a particular medication for his condition. He had watched a video of a prominent political figure endorsing this medication as effective and safe. Naturally, my suspicions were raised, but for a person who is unaware of what AI can achieve, he could have ended up thinking that the video was genuine, as the AI-generated voice in the video was indistinguishable from that of the actual person.
AI literacy is already being incorporated into primary and secondary school education. While the younger generation of doctors is unlikely to need persuasion to use AI, the older generation may be slow to adopt it. Ultimately, the responsibility of patient care still rests with the individual clinician, not the AI. Nonetheless, we need to move forward with the times and learn how to harness AI correctly.