India has a massive shortage of doctors – only 7.2 per 10,000 people, as against 7.6 in Myanmar – a country racked by civil war. Even Pakistan has 11.6, and Cuba – under US embargo since 1960 – has 95.4. Finding a doctor in rural India is much harder as the average plummets to 3 per 10,000, the same as Afghanistan. How many decades will it take to boost India’s doctor strength to, say, Brazil’s 23.6?
If the answer is 10 years or more, the question arises, will we need that many doctors then? That’s because AI is already showing promise as a doctor’s assistant, and in 10 more years it could well be “the doctor”. Not a neurosurgeon, but your friendly neighbourhood go-to physician. Robert Wachter , chair of University of California’s Department of Medicine, discusses this possibility in detail in his book A Giant Leap.
Wachter says this is not the first time AI has been pitched as a substitute for human doctors. Attempts were made in the 1970s too, but the technology then was so underwhelming, and the goals so ambitious – they tried to use it for diagnosis – that failure was inevitable. Then, in the last decade, when IBM’s Watson supercomputer beat humans at a game of ‘Jeopardy’, healthcare seemed like an achievable goal. IBM invested billions, only to dump the project after its bot “prescribed a cancer patient with bleeding be given a cancer medication that can cause severe haemorrhage”.
Yet, three years after ChatGPT’s launch, Wachter and many other medical experts are less sceptical. Wachter reminds you that while AI today is already impressive, “it is the worst you will ever see.” Which means things will only get better from here. Sensing an opportunity, hundreds of startups have entered the field, and some, like Hippocratic AI, started by Indian-origin Munjal Shah, already seem to have a business plan.
Because AI excels at learning from patterns, give it millions of patient files to learn from, and it will correlate symptoms to likely causes: “One study found it could pass a key medical student licensing exam,” says Wachter. On another difficult test, it “performed at a level equal to the best faculty diagnosticians”. So, AI can “think” like a doctor, but can it show empathy like a human? Wachter points to a 2023 study in which AI’s responses were judged to be more empathetic than those given by actual physicians.
So, why hasn’t AI disrupted medicine already? The motivation to do so is very strong when America is spending 20% of its GDP on healthcare, leaving relatively little for infra, and India is struggling with rising healthcare costs and doctor shortages. Wachter says it’s because of problems like bias and hallucinations that AI is susceptible to. One wrong diagnosis could mean curtains for a startup. That’s why, for now, AI is being used for simpler but time-consuming tasks like patient record summarisation.
The holy grail, of course, is diagnosis, but that won’t happen until companies are 100% confident about their bots’ capabilities. For the foreseeable future, AI will only be used to reduce doctors’ bureaucratic burden – documentation, etc – so that they have more time for patients. As for replacing doctors, it won’t happen anytime soon, although roles like radiologist and pathologist could start disappearing in 10 years.
If the answer is 10 years or more, the question arises, will we need that many doctors then? That’s because AI is already showing promise as a doctor’s assistant, and in 10 more years it could well be “the doctor”. Not a neurosurgeon, but your friendly neighbourhood go-to physician. Robert Wachter , chair of University of California’s Department of Medicine, discusses this possibility in detail in his book A Giant Leap.
Wachter says this is not the first time AI has been pitched as a substitute for human doctors. Attempts were made in the 1970s too, but the technology then was so underwhelming, and the goals so ambitious – they tried to use it for diagnosis – that failure was inevitable. Then, in the last decade, when IBM’s Watson supercomputer beat humans at a game of ‘Jeopardy’, healthcare seemed like an achievable goal. IBM invested billions, only to dump the project after its bot “prescribed a cancer patient with bleeding be given a cancer medication that can cause severe haemorrhage”.
Yet, three years after ChatGPT’s launch, Wachter and many other medical experts are less sceptical. Wachter reminds you that while AI today is already impressive, “it is the worst you will ever see.” Which means things will only get better from here. Sensing an opportunity, hundreds of startups have entered the field, and some, like Hippocratic AI, started by Indian-origin Munjal Shah, already seem to have a business plan.
Because AI excels at learning from patterns, give it millions of patient files to learn from, and it will correlate symptoms to likely causes: “One study found it could pass a key medical student licensing exam,” says Wachter. On another difficult test, it “performed at a level equal to the best faculty diagnosticians”. So, AI can “think” like a doctor, but can it show empathy like a human? Wachter points to a 2023 study in which AI’s responses were judged to be more empathetic than those given by actual physicians.
So, why hasn’t AI disrupted medicine already? The motivation to do so is very strong when America is spending 20% of its GDP on healthcare, leaving relatively little for infra, and India is struggling with rising healthcare costs and doctor shortages. Wachter says it’s because of problems like bias and hallucinations that AI is susceptible to. One wrong diagnosis could mean curtains for a startup. That’s why, for now, AI is being used for simpler but time-consuming tasks like patient record summarisation.
The holy grail, of course, is diagnosis, but that won’t happen until companies are 100% confident about their bots’ capabilities. For the foreseeable future, AI will only be used to reduce doctors’ bureaucratic burden – documentation, etc – so that they have more time for patients. As for replacing doctors, it won’t happen anytime soon, although roles like radiologist and pathologist could start disappearing in 10 years.
You may also like
War abroad, turmoil within: Pakistan's costly gamble against Afghanistan
Smart Savings for Your Child's Future: Comparing NPS Vatsalya, Sukanya Samriddhi Yojana, and PPF
Near total shutdown in Telangana over reservation for Backward Classes
Indian Listed New-Age Tech Company Tracker: Market Cap, Revenue & More
Bank FD Loan vs Personal Loan: Which Option Saves You More Money and Offers Better Flexibility?
