I spent six years at UCD medical school in Dublin to become a doctor followed by four years in hospital specialities and then into GP training finally qualifying as a GP after 11 years of study. I remember going into one of my final exams on paediatrics thinking that my brain wasn’t big enough to hold all the information I was expected to know. Of course accumulating information was the least part of my training as the next steps of mastering the skills required followed by the many years of experience needed was a greater challenge.

The huge increase in medical knowledge, not to mention the expanding list of drugs and therapies available are too great now for any human brain to retain which is why artificial intelligence (AI) will come to dominate healthcare.

One of the ways in which the medical profession has managed this complexity up to this point is to have discrete specialists. GPs are the generalists like car mechanics who can do most things with hospital specialists working on particular areas, some based on anatomy; liver, heart, lung, kidney specialists and some based on bodily function or age such as psychiatry, diabetes, paediatrics, and geriatrics.

It is the GP‘s job to focus on prevention, vaccinating children so that they don’t die from measles for example or diagnosing and treating blood pressure so that heart attacks or strokes are prevented. Hospitals tend to work on patients when their bodies break down or are damaged whereas GPs work on the “car service“ element of trying to avoid breakdowns.

We are now told that driverless cars have few accidents because AI drives cars more safely than humans. Within the next decade we will all have our own AI avatar monitoring and advising us on our healthcare like a benevolent granny. However in the transition to that stage we now have clever young patients coming into GP surgery with a particular symptom, for example a pain in the side with an AI differential diagnosis of all the diseases that might be causing it, the list of blood tests, x-rays and scans that ChatGPT AI has advised they need and a list of the various medications that might help.

These patients are genuinely worried and they’ve been completely stressed by the AI technology that they have looked to for information and advice. The differential diagnosis list always seems to include rare catastrophic conditions which would require multiple tests, scans and referrals to a number of different specialists to exclude.

Routine problems with a common diagnosis and simple treatments are no longer accepted. This will put additional demands and pressure on the NHS which has been set up for what patients need not what they want and demand. This will block access for other patients.

AI is at present like the newly minted doctor that I was when I left UCD medical school in 1984, full of information but with limited skills and judgement and no experience, requiring further training.

GPs are the generalists like car mechanics who can do most things.

I have no doubt that AI healthcare will improve rapidly but in the meantime patients might discover that too much medical intervention is actually bad for you and will need to accept that common conditions are common and that their tiredness is unlikely to be Rocky Mountain Spotted Fever!