A technological prescription
Doctors have to be more aware of technological progress
Digital devices are clinical confession boxes. When we feel under the weather or worried about a health problem, we use our phones, tablets and computers to seek medical advice. Surveys show four in ten Americans use Google instead of seeing a doctor, and more than half of Europeans hunt for health information online.
Medical schools are training students to be analogue doctors
In these quests for answers to our medical problems, search engines like Google spit out pages of websites. Now a new generation of chatbots — dependent on what are dubbed “large language models” — can enter into a form of dialogue with us. OpenAI’s “ChatGPT” hit the news in November 2022, and it has scarcely been out of the health headlines since. A bit like predictive texting, this type of chatbot draws on vast amounts of text to generate its responses. It can even “remember” previous questions we asked, helping to build the sense of a tête-à-tête with technology. These chatbots are seamless to use, and they come across as authoritative and humanlike.
Little wonder that the world’s leading medical journals are bulging at their cyber seams. Numerous articles focus on how ChatGPT and other machine learning innovations could be poised to affect the delivery of healthcare.
With all this attention, you’d imagine physicians would, figuratively speaking, have their fingers on the pulses of the bots. Sadly not. Even more worryingly, clinicians don’t seem very capable of reading “machine learning” vitals either. This year, the Ipsos Digital Doctor global survey — one of the largest annual surveys of its kind — reported that most doctors are excited about AI, but only one in three feel knowledgeable about it. Other studies, including those I’ve conducted, demonstrate similar results. Together with colleagues at Harvard Medical School, we surveyed UK GPs for their opinions about the role of AI in their jobs and found most seemed ill-prepared for the digital age.
There are good reasons for the lack of digital savvy. Most doctors are too busy with their day job to stop for breath, never mind keep up with the bots.
We might therefore pin our hopes on the younger docs — the health professionals of tomorrow. Despite small pockets of activity in elite teaching hospitals, the picture is just as dispiriting: AI education is not routine. Consider Ireland, a country with the fastest-growing tech workforce in Europe. My team surveyed final-year medical students there to explore AI exposure during their studies. More than two in three told us they spent zero hours discussing artificial intelligence throughout their entire degree program. A staggering four in ten confessed they never even heard of the term “machine learning”, and 80 per cent admitted they had never read an academic article on AI. Yet most conceded tech advances should be part of their training.
Medical schools are training students to be analogue doctors with education stuck in a 20th century rut. We need to look no further than medical curricula to help explain digital deficits. Despite recommendations for medical curricular reform in the UK laid out in 2019 in the “Topol Report”, there is still a failure of digital literacy across British medical schools. In its “outcomes for graduates” guidance, the General Medical Council omits any mention of the words “artificial intelligence” or “machine learning”. The same is true of North America. In March 2023, in the very latest curricular specifications described by the Liaison Committee on Medical Education — the body that outlines the standards for medical degrees awarded in Canada and the United States — education about AI is wholly missing.
Algorithmic biases could lead to worse discrimination in healthcare
Doctors need to wake up and smell the silicon. The imperative for medics to keep up to date has never been clearer. Health systems are creaking, as any of us who have recently tried to access a doctor know well. COVID-19 oversaw the flight of physicians from primary care. Globally, the World Health Organization predicts a worldwide shortage of 10 million healthcare workers by 2030. Combined with ageing populations and more people suffering chronic illness — and for longer — it is not just clinicians but entire health organisations feeling the strain.
Chatbots like GPT-4 could well prove to be game changers. They are unlikely to fully replace doctors. Rather, AI tools could function as the equivalent of medical-sous-chefs, assisting physicians with a wide variety of tasks. Physicians are drowning in paperwork. ChatGPT could help with writing documentation — one of the biggest leeches on their time. Bots might even offer a bigger shoulder to cry on than clinicians. In one blinded study, a panel of doctors judged ChatGPT to be more consistently empathetic in answering patient problems than other physicians. Chatbots are super swots too, passing medical licensing exams. They can even aid clinicians with complex problem-solving. Preliminary studies show bots helping doctors brainstorm by coming up with diagnoses they might have failed to consider.
Despite all their powerful promise, however, current technology is no panacea. These tools carry real clinical risks too, and not just in exposing patient information to tech giants. So-called algorithmic biases that could lead to worse racial, ageist or gender discrimination in healthcare, not to mention increased risk of medical errors (ChatGPT tends to make stuff up). A host of healthcare hells can arise with these innovations, which is exactly why doctors ought to demonstrate leadership.
Instead of turning a blind eye to the bots, medicine must develop clinicians with critical aptitude. We need a new generation of doctors who know when AI tools are genuinely helpful and when these devices can cause real harm.
To tame technology and get the best from it, doctors and medical schools will have to do a better job of engaging in the modern machine era. If patients are to turn to doctors instead of smartphones, they need clinicians capable of commanding AI. In the long run, patients need medics who fully grasp their rising competition.
Enjoying The Critic online? It's even better in print
Try five issues of Britain’s newest magazine for £10Subscribe