February 25, 2026
A talk I listened to on how AI works with healthcare left me puzzled. A PhD working at the hospital explained LLMs (Large Language Models), such as Co-Pilot, ChatGPT, and Gemini, and compared them to search engines like Google and Bing. I thought I understood the difference: while search engines provide sources for queries, such as links or URLs, LLMs combine multiple sources into text and provide a human-like form of written communication.
According to the speaker, search engines maintain up-to-date information by constantly crawling the web, while LLMs remain current only up to the time they were trained, that is, the information they were given. This is so because not all LLMs are tied to the internet, so they do not have up-to-date information. So, I took it that search engines are good for research, while LLMs offer text and context to a query.
Both tools share the same starting point: the prompt. A question or prompt to a search engine would bring up links and URLs, while an LLM would generate a complete essay on the subject.
I was aware of much of this information; I have used search engines for years, and in the last year, I tried my hand at using some LLMs, for example, Co-Pilot, which is built into the MS Office suite. But how does all this technology relate to healthcare? That was the title of the presentation. So, I asked for an example at the end of the lecture. I said I see a cardiologist sitting at a computer, an MRI image on the screen, and a patient on the side. What does he do to use AI? Does he use a prompt about the MRI taken of the patient sitting next to him? Instead of a direct answer, the speaker explained that the infrastructure for using AI is not yet complete; for example, the hospital must have its own storage for the data, it cannot use storage elsewhere due to privacy concerns.
Also, I did not find answers to how doctors are trained to use AI or what percentage of them are currently using it. Anecdotally, I remember that when I last visited my ophthalmologist, he was concerned and dubious about computerizing his patients’ information. I wonder how long it may take for the medical profession to embrace and use computer technology, especially AI.
Coming back to a live example of how the medical profession may use AI, would he/she ask AI to examine the MRI and identify any medical problem shown on the image? The speaker emphasized that AI is only a tool to use and ultimately, the doctor decides on treatment, potentially using information gleaned with AI tools. After the lecture, I told the speaker that a few examples of medical professionals using AI would have been useful.
To test what I learned, I came home and used a prompt: “Write 300 words on the Vatican Museum”, and asked Google and Co-Pilot, to respond. To my surprise, both apps wrote a nice essay. It appears that the boundaries between search engines and LLMs are blurring, especially as technology advances and these tools integrate.
However, while both can provide information, their use in the specialized, privacy-sensitive field of healthcare remains challenging. There is potential, but also barriers in infrastructure, training, and trust.
PS. In today’s Washington Post, February 24, there is an article by a doctor about how to use AI in medical practice. “AI didn’t replace me as a doctor. It made me better.” By Ashish K. Jha
https://www.washingtonpost.com/opinions/2026/02/24/health-care-ai-doctors-chatbots/
andrasthehun.ca 