You may have already seen it without knowing. The radiologist’s screen shows a chest X-ray with a red box drawn around one corner. The doctor at the desk speaks into a microphone instead of writing, and a typed summary appears. Somewhere in the building, software is quietly sorting the queue.<\/p>\n
Artificial intelligence has arrived in Indian hospitals faster than most patients realise. This article explains what it is genuinely doing, what it is not allowed to do, where it goes wrong, and what you can reasonably ask your own doctor.<\/p>\n
Forget the film version. In practice it is doing a handful of narrow, boring, useful things.<\/p>\n
Reading scans as a second pair of eyes.<\/strong> Software marks up chest X-rays for signs of tuberculosis, flags a possible bleed on a head CT so that case jumps the queue, or picks out a suspicious area on a mammogram. A doctor still looks and still decides. The software is a prompt, not a verdict.<\/p>\n Screening large numbers of people quickly.<\/strong> This is where it has made the most difference in India. Diabetic eye disease is a good example: a camera photographs the retina, software sorts the images, and the small proportion that look abnormal go to an eye specialist. That lets one specialist cover far more people than they could see individually.<\/p>\n Tuberculosis.<\/strong> India’s National TB Elimination Programme has been folding AI-assisted screening into its work, and a reported 27% fall in poor TB outcomes has been associated with it. TB is a disease where finding cases early changes everything, so speed here is not a small win. Our tuberculosis page<\/a> explains the disease itself and what treatment is free.<\/p>\n Paperwork.<\/strong> Less glamorous and possibly more important. Software that transcribes a consultation and drafts the notes gives the doctor back time that goes to patients.<\/p>\n Watching patients in intensive care.<\/strong> Monitors that spot a pattern in the numbers a few hours before a human would, and alert the nurse.<\/p>\n The honest answer is: for these narrow tasks, mostly yes, and Indian clinicians are more positive about it than their counterparts elsewhere.<\/p>\n In a 2026 survey, 71% of Indian healthcare professionals said AI had increased the number of patients they could see — against about 50% globally — with those reporting a gain seeing a median of ten more patients a week. 82% reported better workflow efficiency, and 80% said it improved their access to a patient’s information gathered in one place.<\/p>\n Ten more patients a week matters in a country with this doctor-to-patient ratio. If AI mainly buys clinicians time, that is a real public health gain even if nothing about it is exciting.<\/p>\n This part matters more than the last one, so it is worth being blunt.<\/p>\n It is not deciding your treatment.<\/strong> No AI system in India is licensed to diagnose you or prescribe for you. A qualified doctor signs off, and the responsibility is theirs. If anyone tells you the computer decided, that is not how it is supposed to work.<\/p>\n It is not replacing your doctor.<\/strong> The tasks it does well are pattern-spotting on images and text. It does not examine you, take a history, notice that you look unwell in a way the numbers do not show, or judge whether treatment is worth it for you specifically.<\/p>\n It is not neutral about who you are.<\/strong> A system trained largely on scans from other countries may perform differently on Indian patients. Good hospitals test tools on their own population before trusting them; not every hospital does.<\/p>\n It can be confidently wrong.<\/strong> This is the failure mode people underestimate. AI does not hesitate or hedge the way an unsure human does. A wrong answer arrives with exactly the same calm certainty as a right one. Chat-style AI in particular can invent facts, references and drug names that look entirely plausible.<\/p>\n People trust it too much.<\/strong> There is a well-documented tendency for humans checking a machine to gradually stop really checking. If the software has been right ninety-nine times, the hundredth gets less attention — and that is the one that mattered.<\/p>\n Your data goes somewhere.<\/strong> Your scan or your notes may be processed on servers outside the hospital. India’s Digital Personal Data Protection Act, 2023 gives you rights over your personal data, and you are entitled to ask what happens to yours.<\/p>\n It is uneven.<\/strong> A large corporate hospital in a metro and a district hospital are not running the same tools with the same oversight. “The hospital uses AI” tells you very little on its own.<\/p>\n None of these are rude. A good doctor will not mind any of them.<\/p>\nIs it working?<\/h2>\n
What AI is not doing<\/h2>\n
Where it goes wrong<\/h2>\n
What you can ask your doctor<\/h2>\n
\n