Will Artificial Intelligence Replace Doctors?

It is a fact commonly known that advances in technology displace people from their work. While victims have mostly been menial workers, artificial intelligence – AI – promises to be the great equaliser, coming to lay waste to fields that have hitherto withstood the technology challenge. The question is – is this going to apply to doctors too?

This is a topic that appears to be hotly debated wherever it occurs

Two years ago I opined in one forum that technology was going to make doctors redundant in the not so far future. The doctors who were present were not amused. The idea appeared quite sacrilegious to them.

Being a doctor is a highly exalted profession, a profession that deserves everyone’s respect. Therefore, the challenge that we face when analysing the impact of new technologies on the practice of medicine is not to think too highly of doctors – we already do that – but to think too little of technology. Many of us haven’t had the exposure needed to appreciate how mind-boggling modern technology truly is.

The arguments that doctors usually give in defence of their profession are threefold. One, machines work in a linear patterns, so they are not well equipped at diagnosing complex conditions. Two, machines cannot match humans’ dexterity and capability in performing intricate surgeries. Three, machines can’t understand human psychology.

Let’s examine these arguments one by one.

Doctors, Diseases and Diagnoses

Firstly, we need to understand how doctors diagnose diseases. Doctors use ‘profiles’ of diseases to diagnose conditions. One disease profile will often overlap with another; thus, a diagnosis is usually a probability that is laden with uncertainty. This uncertainty is not bad: when we see heavy clouds, we reasonably conclude that it is going to rain. The problem is – sometimes it doesn’t. And this is the problem with the doctors’ method.

Machines, alternatively, only work with evidence. If you go to a clinical or imaging laboratory where your sample will be analysed for 100s of parameters, machines, by determining correlations between hundreds of parameters to deduce a diagnosis, a process that requires millions of computations, will always produce better diagnoses, thus removing the need to be creative. Therefore, from diagnoses of skin, lungs, or breast cancers, to those of heart conditions and retinal issues, reports of machines outperforming doctors abound. In fact, machines don’t even need patients’ histories or to be told what to look for – they outdo humans because they are better at dealing with complex data.

But, 100s of parameters, who is going to pay for that? That is now a question for economics. The point here is that doctors are only useful in this equation to optimise the number of tests that are to be done, not to add value to the clinical process. Alas, technology will make it more economical to do more tests per visit with time.

Machines are Already Outperforming Experts

Secondly, surgery. The field of robotics has undergone dramatic improvements lately, hence we will soon move from guided robots that assist doctors in surgeries – say, in brain surgeries for precision or for quality of their sutures – to complex surgeries which require significant repetitions for mastery, such as heart and pancreatic surgeries. Unlike doctors, machines don’t need dozens of operations to acquire mastery, so, machines will be preferred since they can be mass produced and get deployed - initially in places where certain skills are absent - then come back to places where surgeons are and make them redundant.

Machines and Empathy

Thirdly, psychology. It is usually argued that machines don’t have empathy, a necessary component in care giving. Indeed, we don’t expect robots to ‘feel’ as we do, but it is wrong to think that machines cannot get human psychology. Actually, it is now possible to diagnose multiple conditions such as sociopathy, depression, and stress with expert accuracy. Machines are already helping doctors with proactive management of cases such as bipolar disorder and suicidal inclinations and, in the UK, machines are deployed to give medical advice – again – with expert accuracy.

The idea that humans are better at being human than machines is misguided. AI is a truly game-changing technology: this is the technology that can improve itself.

A machine recently produced an aeroplane wing in a way that no engineer could envision – halving the weight while increasing the strength. A Google’s Alphago machine beat a Chinese master of a complex board game ‘Go’ – not even its own makers understood its moves. These machines can be taught to mimic human emotions and, indeed, given that there are humanoid robots today which can hardly be distinguished from their makers, it is only a matter of time before we feel more comfortable in their hands than in our ‘temperamental’ human beings.

Final Thought

The fact is that when it becomes cheaper and more effective to deploy a machine than a doctor, doctors will become as dispensable as other workers, and medicine will be driven by tech entrepreneurs rather than doctors and scientists. Welcome to the future.

This article first appeared on the Citizen newspaper. We would like to hear your views. Please like, comment, and share as needed. Cheers.