Debate: Will AI replace clinicians?

As part of the ongoing conversation of artificial intelligence in healthcare, two GPs debate whether or not AI is a threat to the workforce

YES: ‘AI will replace clinicians and reduce the workforce to 10% of what it currently is’

Like it or not, artificial intelligence is becoming an increasingly dominant force in healthcare, and one that is getting harder and harder to escape. Many of us probably already use it in our day-to-day work: helping to reduce administrative loads (coding letters and recording consultations); helping with clinical decision-making (patient triage or prescription tools); or assisting patients (wearable health technology and AI-powered home testing kits.) 

But there still seems to be a reluctance among practitioners to fully embrace the technology – and for good reason. I believe that AI will replace clinicians and reduce the workforce to 10% of what it currently is. We are approaching that point at a rapid pace already. In March, American tech company NVIDIA partnered with AI healthcare start-up Hippocratic to create ‘AI-generated nurses.’ Closer to home, last year ChatGPT – a non-healthcare specific machine – scored 96.7% on the AKT exam. 

Critics will always point to clinicians’ ‘gut instinct’ giving them the edge over AI tools, but gut instinct is nothing more than years of an individual’s experience. If you feed a collective knowledge/experience into technology, AI will be a far more valuable asset than one singular clinician’s ‘gut instinct’ could ever be. 

Regarding the practicalities of patient treatment and examination, AI will be well-equipped to deal with both mental and physical consultations. 

For example, the principles of CBT involve an evidence-based appraisal of your thoughts which subsequently lead to your feelings and behaviour. It is a highly scientific approach, and we already have computerised CBT in use for patients. It is a small step to progress this to AI driven psychotherapy. 

Regarding physical examinations, AI also wins. It will be far better than me at identifying whether a mole has become cancerous. It can be used, not only for analysing photographic images, but current technology can actually visualise its user and comment on their appearance. This advance certainly has benefits for physical examination and will be augmented through the development of robotic appendages to examine a patient. That said, it is well accepted that over 80% of a diagnosis is in the history, so we are only trying to reach the last 20% of the puzzle.

Of course, the most common argument against AI is its emotional capacity. But with new AI tools on the market which can detect and deliver sarcasm and humour, it’s a short step away from emulating the much needed empathy patients need. Whether it can replace the tender loving care which comes from holding a patient’s hand through their journey, is harder to tell, but nothing surprises me with the speed of advancement.  

As a final point, it is worth considering, not just if AI has the technological capabilities to replace clinicians, but whether or not the prospect will be embraced within given political, economic and social climates. Because we have a nationalised healthcare system in the UK, there is limited funding. And in such a system like the NHS, having a human clinician – a doctor – is a luxury. This ‘luxury’ is increasingly becoming something that we can’t afford and if AI can replace humans for a cost-effective solution, then you can guarantee it is a decision that the higher-ups will not hesitate to make.

I see the attrition of the workforce growing exponentially in the very near future. You’re going to have – I would say – 90% of the work we currently do be powered by AI soon. The priority then will be making sure that the clinicians who do remain have their terms and conditions greatly improved so that they are retained. If it is only 10% of our current workforce, then that is probably affordable then. 

AI will absolutely replace clinicians. It already is.

Dr Shaba Nabi is a GP trainer in Bristol and a former Pulse columnist. Her son, Laith Grice, also contributed to this piece.

NO: ‘Healthcare is fundamentally a people industry’

In healthcare, as in life, the only constant is change. Throughout the history of modern medicine, clinicians’ roles have continually evolved and morphed in step with technological advancements. New specialities have emerged, roles redefined and work redesigned. 

These transitions have always resulted in redundancy; not of clinicians, but rather a redundancy of tasks. Interventional radiology has reduced the need for surgical procedures. Automatic sphygmomanometers have reduced the need for manual blood pressure measurement. E-prescribing systems have taken over from hand-written drug charts. Telemedicine has reduced the frequency of face-to-face consultations. Technology has rendered some tasks obsolete, but not clinicians. 

Therefore, surely a more relevant question than ‘will AI replace clinicians’ jobs’ is ‘which of a clinician’s jobs will be replaced by AI?’

We are in the midst of a global healthcare workforce crisis with an estimated shortage of 10 million health workers by 2030. The healthcare workforce is also facing record levels of demoralisation and burnout. Ultimately, AI cannot replace a workforce that doesn’t exist. However, if implemented astutely, AI could help to plug the workforce gap and help to blunt burnout. 

Administrative tasks will likely be the first domain of AI task redundancy in medicine. Us clinicians know that we spend around half our time working on computers. AI automation could open up the potential of ‘keyboard liberation’ in the consulting room. Several AI tools are already being used in healthcare that can automatically transcribe and summarise consultations, write clinical notes, and create referral letters. Such tools could reduce time pressure and cognitive load on clinicians, therefore allowing them to spend more time on meaningful clinical encounters with patients.

Technological advancements are a catalyst for the creation of new roles and novel clinical specialisms. The development of safe anaesthetics significantly increased demand for surgeons, produced the speciality of anaesthetics, and led to the creation of new roles such as operating department practitioners (ODPs). Radiation breakthroughs led to the emergence of the radiology speciality and radiographers entered the workforce. Genetic screening opened up clinical genetics as a medical speciality and resulted in the employment of genetic counsellors. 

We can expect to see a similar pattern of job and role creation with the introduction of AI technology. An influx of data, digital, and technology professionals – and a new breed of AI-literate clinicians – will be needed to implement, critically appraise, govern, and create effective clinical AI. We can anticipate the formation of multidisciplinary AI teams, AI specialist clinicians and the emergence of senior digital and AI positions within healthcare.

Geoffrey Hinton, one of the ‘godfathers of AI,’ came under fire in 2015 when he claimed that within five years radiologists would be replaced by AI. Nearly a decade later, the demand for radiologists is higher than ever and the doctor-patient relationship built on trust, empathy and confidentiality remains the cornerstone of medical practice. 

Human touch, laying on of hands, active listening, and the intimacy of human connection; there are many aspects of healing in medicine that have little to do with a replicable clinical skill or capability. Healthcare is fundamentally a people industry and people will always play a central part in delivering it. However, that doesn’t mean the tasks those people do won’t change. The biggest threat to clinicians’ jobs is not AI, but clinicians who can use AI. I’d recommend becoming one of them.

Dr Anabelle Painter is a GP registrar and honorary digital health fellow at Imperial College