All the months’ work in today’s time could be done by a computer in few minutes. MUMBAI: When I was a student of medicine, I was once asked by my chief to get a list of references on the subject of ‘mycoplasma’. I got busy with this – in those days, one had to look up the Index Medicus, a huge compilation of listed research, to chase this. It was over a month working several hours a day that I could finally arrive at a comprehensive list and my chief was pleased.
An interesting article about AI in the New England Journal of Medicine’s (NEJM) 2026 annual issue by A Sekora, L A Ceti and Abdul Nour addresses the drawbacks of AI. During the Covid-19 epidemic, many healthcare organisations around the world tested new AI-based tools to monitor patients with Covid. Humans can identify homosexual faces 61% of the time with males and 54% of the time with females.
AI has tremendous potential to improve patient care – its importance is particularly paramount in radiology where one can also avoid human error in interpretations of images. It states that clinicians are expected to disclose their gaps in knowledge, but emerging AI tools cannot or will not do the same. This is correct. Whenever I am confronted by a confusing set of symptoms, instead of ordering random investigations, I tell the patient I need to think over the problem or even discuss with a colleague about the way forward. I remember one of my professors in medical college who repeatedly reminded me that the uncommon manifestation of common disease is more common than the common manifestation of uncommon disease. The NEJM article states that emerging AI tools cannot say or will not say the same. This, in my opinion, may lead to using AI-based algorithms down the garden path, often ordering useless expensive investigations. The Harvard Gazette states that AI can provide evidence-based medical literature and support real time decision-making at the point of care. The results of these exercises are still being evaluated and it is likely that AI will become a part of digital health system that defines modern medicine. AI technologies can also be used to determine sexual orientation. An article published in The Economist tells us that research by Michal Kosinski and Yilun Wang, from Stanford University, shows how sexual orientation can be somewhat accurately determined by AI. AI system relies on eyes, eye brows, cheeks, hairline and chin to evaluate male sexuality and the nose, corners of the mouth, hair and neckline to evaluate female sexuality. Perhaps the human brain relies on such features as well. In essence, AI technology certainly helps with radiological imaging. As far as the practice of diagnostic medicine goes, it will surely be a useful tool. The wise physician relies not only on databases of evidence, but data bases of experience.
