Some AI experts opine that AI will completely take over the practice of medicine in 7 years. More conservative opinions put that at 20 years.
Physicians invariably trigger denial mode and say, it couldn’t possibly do my job. But if you dissect piece by piece what they think AI can’t do, turns out, AI can or will do it faster, better, and cheaper.
One EM Doc said: AI can never diagnose a psychotic patient or perform and dislocated joint reduction. With a compendium of 100,000’s of joint reductions in its database – and the inability to tire, joint reductions for AI and robots would be almost effortless. Noticing the facial movements, vocal rhythm and tempo, word choice, AI could diagnose a psychotic patient in minutes if not seconds.
Another doc said, AI will never be able to tell shortness of breath from a PE from the patient who says they’re short of breath but means they can’t breath through their nose from a URI or sinusitis. This person clearly underestimates AI.
Just two weeks ago the Lancet had a huge study showing AI alone diagnosed breast cancer 29% better than radiologists with or without AI – and with NO increased false positive rate.
The data is so overwhelming that one might say it’s malpractice not to use AI in these *specific* areas with this degree of evidence. Would you want yourself or your family member not to have a 29% improved breast detection rate on your screening?
Those who invariably say our jobs are safe, are basing their sense off of linear patterns of improvement.
The issue? AI is advancing exponentially – at 10x per year. And the pace is not slowing, it’s hastening!
AI never gets tired. It can diagnose 1M unusual illnesses and come back for more. And AI today is the least capable it will ever be. In 3 months it will be twice better. That pace of improvement is beyond insane.
There are tens of companies in a race to develop capable robots hand-in-hand with the advancement of AI. The race is not slowing down.
It will undoubtedly converge sometime in the next several years.
One Response to “AI May Allow Physicians To Regain Their Humanity”
February 23, 2025
DocGRSome AI experts opine that AI will completely take over the practice of medicine in 7 years. More conservative opinions put that at 20 years.
Physicians invariably trigger denial mode and say, it couldn’t possibly do my job. But if you dissect piece by piece what they think AI can’t do, turns out, AI can or will do it faster, better, and cheaper.
One EM Doc said: AI can never diagnose a psychotic patient or perform and dislocated joint reduction. With a compendium of 100,000’s of joint reductions in its database – and the inability to tire, joint reductions for AI and robots would be almost effortless. Noticing the facial movements, vocal rhythm and tempo, word choice, AI could diagnose a psychotic patient in minutes if not seconds.
Another doc said, AI will never be able to tell shortness of breath from a PE from the patient who says they’re short of breath but means they can’t breath through their nose from a URI or sinusitis. This person clearly underestimates AI.
Just two weeks ago the Lancet had a huge study showing AI alone diagnosed breast cancer 29% better than radiologists with or without AI – and with NO increased false positive rate.
The data is so overwhelming that one might say it’s malpractice not to use AI in these *specific* areas with this degree of evidence. Would you want yourself or your family member not to have a 29% improved breast detection rate on your screening?
Those who invariably say our jobs are safe, are basing their sense off of linear patterns of improvement.
The issue? AI is advancing exponentially – at 10x per year. And the pace is not slowing, it’s hastening!
AI never gets tired. It can diagnose 1M unusual illnesses and come back for more. And AI today is the least capable it will ever be. In 3 months it will be twice better. That pace of improvement is beyond insane.
There are tens of companies in a race to develop capable robots hand-in-hand with the advancement of AI. The race is not slowing down.
It will undoubtedly converge sometime in the next several years.