As we’re consistently told we must brace ourselves to pay ever more tax to pay for that Wonder of the World that is the National Health service. It needs real budget rises – faster not just than inflation but than the rate of GDP growth – off into the future for decades. At which rate it will swallow the entire economy of course. We’ve just had a £20 billion promise as downpayment and don’t, for a moment, think that it’s going to stop there.
The underlying problem is the following:
The underlying logic here is that of William Baumol, best known for Baumol’s Cost Disease. Average wages in an economy are determined by average productivity. And it is easier to increase productivity in manufacturing than it is in services – the canonical example being that a Mozart string quartet takes the same time to play today by the same number of musicians as it did in the 18th Century, when he wrote it.
Given the labour embedded in each, services will become more expensive relative to manufactures over time. Another example is that a custard tart, from an automated production line, would be far cheaper than the one prepared by the hand of a baker offering a distinctly personal service.
Health care is largely a service. Certainly, the vast majority of the cost is wages. There is a solution to this. Automate the service, turn it into a manufacture, and we can and will increase the productivity. Which is what is happening here:
Claims that a chatbot can diagnose medical conditions as accurately as a GP have sparked a row between the software’s creators and UK doctors.
Babylon, the company behind the NHS GP at Hand app, says its follow-up software achieves medical exam scores that are on-par with human doctors.
Well, obviously the doctors are annoyed. Who would want some mere robot, AI, to disrupt what is a rather well paid profession?
The company said its AI scored 81% when it was tested using RCGP exam questions, whereas the average mark for real-life doctors was 72%, according to data from 2013 to 2018.
However, the RCGP said the claims were ‘dubious’. It highlighted Babylon had used its MRCGP exam preparation questions, which were for ‘revision purposes,’ and that these ‘are not necessarily representative of the full-range of questions and standard used in the actual MRCGP exam’.
The professional body is simply never going to admit that software could do as well as its members now, is it?
“The World Health Organization estimates that there is a shortage of over 5 million doctors globally, leaving more than half the world’s population without access to even the most basic healthcare services,” Babylon founder Dr. Ali Parsa said in a statement. “Even in the richest nations, primary care is becoming increasingly unaffordable and inconvenient, often with waiting times that make it not readily accessible. Babylon’s latest artificial intelligence capabilities show that it is possible for anyone, irrespective of their geography, wealth, or circumstances, to have free access to health advice that is on par with top-rated practicing clinicians.”
And that’s what scares the ordure out of the professional body, isn’t it?
It’s happened before to professions – we’re really, to any useful statistical level, entirely out of sailing masters, in its time a very highly paid and well respected profession – and the usual reason for the disappearance of a profession is that technology automates the task.
Much of doctoring is rote learning then following a decision tree. This is something almost designed for an AI to be able to do well. Maybe this iteration isn’t quite there although IBM has been making interesting claims for Watson. But what’s near inevitable is that an AI is going to do it better than a human and real soon now too.
Which is interesting, because when it does we’re going to have the most almighty dogfight. The outcome of which will be, in the end, a cheaper NHS for us all and possibly a better one. How long it takes will be determined by how good the rearguard action from the RCGP is. Hopefully, really terrible. Look out for an insistence that the AI may only be used by a qualified doctor – the equivalent of the man with the red flag walking in front of a car – for example.
AI can certainly sift out non-diagnoses as well as a physician using that decision tree. (AI in the hands of a government bureaucracy can also pump lots of personal data out of the citizenry, for instant misuse should a skeptic decide to run for office.) For actual treatment, you want a doctor who puts his integrity on the line. It didn’t worry me that all the computers in the bank might malfunction on the morning of Y2K, because my business there rests not on the correct functioning of tools but on the commitment of people to give me my deposit… Read more »
Is it true that it is harder to wring productivity increases out of services than out of manufacturing? Or, if true do we know by what extent. Many services, such as transportation of people and goods have been made much more productive by the advent of motorized transport capable of carrying greater loads or other innovations such as shipping containers. Having a contractor do work on your house is more efficient because of power tools and the ability to pickup materials and supplies at a nearby Home Depot without having to personally cut down a tree and saw it into… Read more »
Tim has pointed out that even the string quartet is more “productive” in making a record versus only playing to live audiences; it satisfies more customers with the same labor.
Indeed. Even they are more productive than in Mozart’s day.
One can at least hope that, unlike certain human GPs, an AI won’t stuff you full of diamorphine until you die.
QV,
Depends on whose medical notes have been used to train the AI. If it learns that cranky old git requires higher doses of diamorphine Tim is going to start losing commentators on his own blog at an alarming rate.