No, Your Doctor Won't be Replaced by a Robot
AI has been taking over medical tasks for 50+ years – improving, not replacing, doctors' ability to care.
My parents moved to the United States in the 1960s as immigrant physicians from Iraq. One of the tasks they regularly did as doctors – taking up ~5% of their time – was to smear collected blood on a microscope slide, then count the white and red cells per high powered field – the complete blood count (CBC).
Today this action has been totally replaced by Artificial Intelligence (AI) – just insert the blood and swish, fizz, whirl… the CBC results are entered into the EHR!
But we don’t have 5% less doctors
Just like any technical advancement, this didn’t eliminate the expert, but rather allowed us to perform a whole lot more CBCs. It has replaced some specific activities - but it has not replaced the physician.
Recently, several large tech companies have tried to sell AI as something new: a mysterious black box that can “sniff out” diagnoses and treatments and eventually replace docs. Beyond their inability to explain how they’re planning to develop such technology - and a general hand waving around “big data” - they seem to ignore that artificial intelligence has been changing medicine for decades. It has not replaced doctors. To the opposition, we are currently experiencing a physician shortage. However, AI has been beneficial on other fronts with its ability to increase the quality and quantity of care we provide.
Software has been developed that can, for example, suggest diagnoses from electrocardiograms, remind us to avoid drug/drug interactions, and automatically interpret monitored pulses and blood pressure. All of these technologies are artificial intelligence: they help make decisions that humans used to make. These resources actually make the doctor more valuable, rather than less, and simply change the expectation of a skilled physician. Doctors in my generation need to know how to use electronic health records and apply evidence-based medicine, not count blood cells on slides.
The same thing is true outside of medicine: customer relationship software, the sales version of AI, does not replace salespeople, but rather increases their output. Microsoft Excel allows bankers to be more accurate in ways no human could before. Smart technology has pushed forward all industries and expanded expectations about what it means to be an expert.
The marketers at tech companies selling ‘AI in Medicine’ know how easy it is for us to confuse the true definition of artificial intelligence with the images it conjures up in our mind - the robots from West World, the car KITT from Knight Rider...
By definition, artificial intelligence is computers helping make decisions human used to make – such as the EKG softwares, pulse monitors, and CBC machines. But “AI” conjures up images of a personified robot that’s hard to distinguish from a human: we think of robots from West World, or the operating system Scarlet Johannsen played in Her, or for Gen Xers like me, the car KITT from the TV show Knight Rider. Marketers exploit this imagery.
Consider IBM’s commercials, which presents Watson (a human name) as an oracle-like black box sitting among people, with a voice and quirky personality. This helps them create a buzz around a product so they never need to really explain or justify what these machines do. This is a bit of a smoke and mirrors show however when you look behind the hype what they are developing is a continuation of the same technological improvement we've seen for 50+ years since my parents trained in medicine.
And for a second, separate reason: physicians will not become robots because society does not want them to.
Doctor comes from the Latin docere, which means “to teach” – to console, to care for, to reassure. 100 years ago - before most medications and surgeries - there wasn’t much else a doctor could do but teach and comfort. Yet this doctoring has been shown to have incredibly powerful and beneficial effects on patients. No computer can replace this.
In this new era of evidence based medicine (EBM), some observers mistakenly conclude that clear measurements will allow computers to make more decisions for physicians. But EBM is a blunt measurement tool, aggregating data across a diverse set of individuals who have significant variability of disease and treatment effect. EBM only works when a practicing physician can apply it to the individual patients in front of him or her.
This humanism cannot be divorced from medicine: not just the empathy when connecting with patients, but also the art of applying the more objective science to individuals who each express disease and pain in their own way.
It’s true that I’m a practicing physician, so one could assume I’m biased against the idea of AI replacing physicians. But I’m also a cofounder of the largest clinical decision support service used by physicians, MDCalc, which provides clinical decision tools to over 50% of US physicians each month. I was a mathematician and statistician before I was a physician. I would greatly benefit – more than nearly anyone – if the practice of medicine could be automated by computers. But I know there are limitations to what can be made objective. The AI of the future will only work when skilled, caring physicians with a deep understanding of evidence-based medicine apply technology to the individual patients in front of them.
About the Author:
Joseph Habboushe, MD MBA, is a practicing emergency physician and Cofounder / CEO of MDCalc, the largest medical reference owned and run by physicians - accessed by over 55% of U.S. doctors per month.
