Daybreak - India is training doctors in AI. Can they build what tech bros can’t?

Episode Date: March 29, 2026

India's hospitals have been slow to adopt AI. Its government, however, has not. A new programme aims to train 50,000 doctors in artificial intelligence. And not just to use it, but to help build it. The argument is simple: engineers understand disease like an algorithm. Doctors know it's never that clean. So what happens when clinicians become co-builders?Tune in. Daybreak is produced from the newsroom of The Ken, India’s first subscriber-only business news platform. Subscribe for more exclusive, deeply-reported, and analytical business stories.

The AI in healthcare boom has been pretty inescapable. Everyone has been talking about it. But what we're also seeing is a very unlikely and enthusiastic adopter. And it's not hospitals. It's the Indian government. The government's plan is pretty ambitious, training 50,000 doctors in AI and making them co-builders. With this, the government is basically trying to reshape how health tech,

is built in the country. But can these doctors do what coders and engineers could not? That's the question my colleague Sudeshna Ray asks in this edition of one of the Ken's most popular subscriber-only newsletter called Make India Competitive Again. Sudeshna's piece is titled, India is training doctors in AI. Can they build what tech grows can't? And today, I'm reading it aloud for you. Welcome to Daybreak, a business podcast from the Ken.

I'm your host, Rachel Virki's, and every day of the week, my co-host, Snigda Sharma and I will bring you one new story that is worth understanding and worth your time. Today is Monday, the 30th of March. The world is a wash in Health AI. Earlier in January, OpenAI launched Chad GPT Health, a feature that can analyze users' health records. Days later, its rival Anthropic launched a similar feature in Clod. Google 2 dipped its toes into this with the release of Med Jemma. Between healthcare infrastructure, consumer needs and clinical potential, AI now covers a gamut. But while hospitals are ignoring this, the Indian government

is catching the AI trend by its haunts, or at least it plans to. On 21st January, the Union Health Ministry launched an AI training program in medical education. It aims to equip 50,000 doctors with digital skills to understand AI applications in health for diagnostics, clinical decision making and research. In fact, even before the six-month program was announced, it had signed on over 42,000 doctors. A healthcare professional associated with the program said that the government is not working in the typical way as far as AI in healthcare. is concerned. They are very systematic and aggressive. The private sector is also enthused. On 30th January, when the Government Institute of Medical Sciences or GIMS in Great Anoida outside

Delhi hosted a startup OPD, bringing together founders and doctors to discuss innovation, several investors also wanted in. Among those eager to hear startup pitches were Acumia Capital, Sunnicon Ventures and Indycon Angels. Each manager, Rupies 100 crore plus funds. They were equally interested in how clinicians at the event evaluated those pitches as well. Rahul Singh Amritraj, who heads GIMS, said that that's the key. That's basically where investors have agreed to come down. The investors said that if your doctors can validate and help us, it becomes easy to invest.

And that wasn't the only AI in health discussion held in the city that day. The Ken learned that a medical regulatory body met with different stakeholders, including doctors and those managing compliance, to formulate regulations for AI use in healthcare. But all of this is just scratching the surface, said Sujit Katiar of Surisulis Ventures, a healthcare strategy and innovation consultancy. The firm helps healthcare enterprises in regulatory compliance. He added that the big picture is to have 50% of the country's total.

health care workforce trained in AI use cases by 2030. Besides 1.4 million doctors in India, that group includes 4.3 million registered nursing personnel, over 1 million ASHA workers who are community health volunteers and 2.4 million Anganwadi workers or frontline functionaries among others. GIMSS Amrit Raj believes that it's important for doctors to be trained in AI and to take part in building modern healthcare solutions. Which is why the Institute debuted India's first government hospital-based AI clinic earlier in January aiming at enhancing patient safety and reducing treatment time.

Additionally, the clinic would incubate a bunch of healthcare startups and test their solutions in daily operations. He also added that engineering folks understand disease like an algorithm. But it's not that way. Disease is very personalized. That is, the solutions developed in the sector. so far have been driven by the kind of profit margins one could gain from them. Satish Rath, a Bangalore-based physician scientist, also agreed. He said that training ensures that doctors solve what needs to be solved.

Those are also ideas that VCs will be eager to fund. Eight months ago, Rath, who has founded several healthcare-focused ventures before, launched GreyBrain, a Gen-AI training course for doctors. It teaches basic use cases of the test. technology, such as managing business processes, collating and interpreting medical data, and even writing clinical papers. But, Rath also said that you can't just ask an applied science to start learning your language. So, AI must be unpacked into the language of doctors.

For instance, the Indian Institute of Health Management Research or IHMR in Bangalore, which has been training doctors in AI for some time now, has developed a foot-size detection tool to help treat leprosy patients. It uses existing data from the doctor's clinical practice. Besides an AI tool deployed at its radiology center, GIMS uses a couple of other applications as well to identify process lags in testing for tuberculosis and early screening of breast cancer. Now this is just breaking the eyes.

Rutz said that with this base, doctors can start imagining use cases and even have their own makeshift prototypes that could resonate with other doctors, startup funders and investors. At the same time, stakeholders are also concerned about the regulatory hurdles this arrangement might run into. With tools like clinical decision support system, which analyzes data and provides treatment recommendations, and Madunitri, which detects diabetic retinopathy, seeping into the healthcare system, assistive diagnosis comes with calculated risks for the clinicians. So, where does the accountability rest then? Amrit Raj said that if the doctor signs a clinical prognosis or diagnosis by an AI tool and then a patient goes to quote, the physician

will be held accountable, not the technology. Which is why training for use is important. Amrit Raj says that you can't promote quackery in AI. Technically, India's new digital personal data protection or DPDP rules place the primary accountability of protecting one's data with the fiduciary, which are the hospitals in this case. That's one of the reasons, said Katyaar, some of them avoid adopting the technology. As a solution, he and his peers are recommending the concept of shared accountability, even with the technology vendors and regulators. Now, globally collecting data from patients was a given, even before AI solutions were on the chart. That's because healthcare systems in most countries abroad are closely tied to insurance.

In India, though, electronic medical records or EMRs are still miles away from being digitized, even though a lot is in the works under the flagship Ayushman-Bharat scheme. Tracking one's full medical history is necessary for the data that AI tools could use. Now, ideas exist for a lot of projects, but data does not, said Akash Prabhune, who is a professor, and in charge of AI medical training at IHMR. In such cases, the Institute uses quantum models where they can get reliable results from limited available data. Usha Manjad, who is the Institute's director,

said that some specialized models like the ones built for oral pre-cancer screening or ulcer detection are built on small data sets. Many don't even have Indian patients' data. That's because oral lesions and tobacco use, users in say Madhya Pradesh are different from those found in users in Karnataka, thanks to deferring cultural habits and the tobacco itself. To understand the clinical aspects of those lesions, one needs to collect data relating to the tobacco source, lifestyle shifts of the patient

and even the geography. Such fragmented data makes it challenging to build an appropriate diagnostic or screening tool. GIMS is said to be currently working with the government and AI-trained doctor peers to develop the India design for AI in healthcare, a model that focuses on the country's specific clinical and healthcare infra gaps. The Institute was also preparing for the India AI Impact Summit, which took place in mid-February where discussions around regulating the tech were expected to come up. That said, more than being AI trained, said Manjouinat, doctors should be digitally literate. She also said that somewhere our education system

created this bifurcation that engineers should code and doctors should treat. And if you look globally, many doctors have technical qualifications and some can even code better than engineers. She added, and I'm quoting here, we have created an illusion that someone with a non-healthcare background should come and build for healthcare. And training doctors in AI is step one of breaking that illusion. Daybreak is produced from the newsroom of the Ken, India's first suburb. subscriber-focused business news platform.

What you're listening to is just a small sample of our subscriber-only offerings. A full subscription offers daily long-form feature stories, newsletters and a whole bunch of premium podcasts. To subscribe, head to the ken.com and click on the red subscribe button on the top of the Ken website. Today's episode was hosted and produced by my colleague Rachel Vargis and edited by Rajiv Sien.