Secondly, AI works on the basis of algorithms, and they need data. In contrast, the whole point about medicine is that it takes seriously incomplete information, the vast majority of it being communicated nonverbally, and interprets that based on the intuitive experience of the practitioner. The weightings provided are those that at the moment when the decision has to be taken (3am at the bedside of an unconscious elderly person with multiple co-morbidities) seem best to the doctor using a combination of all the heuristics that years of experience has provided to them, many of the inputs for which they will never have time to record. It is, in other words, an exercise in the management of risk in the face of extraordinary uncertainty, including very often not knowing what the patient actually has wrong with them.
Good luck in finding any AI system that can process that in a few seconds, including all input time, right now or at any moment in the foreseeable future. It is not going to happen.
That’s exactly what AIs are good at, pattern recognition on the basis of past data and patterns. It’s actually how they work.
One of the first AI systems to beat human experts was the Mycin system to diagnose potential melangitis faster than diagnostic tests could do so , in the mid 70s …
Yes Tim, but
Garbage In Garbage Out
If the data is flawed or incomplete then we get a wrong answer.
That is the real issue with AI.
All it takes is for someone wrongly to identify a cat as a dog and the entire system is bolloxed up.
He’s an expert in medicine now cos his missus is a GP?
Bit like Blackadder.
“ Blackadder: Baldrick, I’ll deal with you later. [to George] Am I to understand that you are going to be representing me at the court martial.
George: Oh, yes, sir. Bit of a family tradition, actually. My uncle’s a lawyer, you know.
Blackadder: Your uncle’s a lawyer, but you’re not?
George: Good lord, no. I’m an absolute duffer at this sort of thing”
Replace George with Spud and it would fit most situations he pontificates on.
Bear in mind we don’t have AI yet and it could be some time away if ever. What we have is Machine Learning with extremely fast hardware, processors, memory and disks and some very smart programmers. That is why when ‘AI’ goes loopy and anti-woke it can be fixed, its simply a modification to the coding.
AI just sounds sexier than ML and is pushed just like SMART devices and honourable politicians. Just words.
The thing with all “AI” is that it works best as a suggestion tool. You enter a title into eBay and it gives you 3 possible categories. It’s not even saying that one of these is right, but that based on the words you entered, and how often those words appear in other categories, it suggests one. And it gets it right a lot of the time, so it saves you a load of typing.
The value of medical AI is mostly about advising a doctor. Not an answer, but mostly likely one of these 5 things. And the real value to me is in things like rare conditions. You distill a ton of historic illnesses and the symptoms, and the doctor thinks “huh. Never heard of that and goes off” and checks. Might be bollocks, but it might get it solved quicker. Or you do less tests to get it to the answer.
AI making deathbed decisions on limited info? I’m sure that will work out fine. Will we be training it on pathways?
Traditional coding works on the basis of algorithms: you have to figure out how to do a task beforehand – the algorithm; a finite sequence of steps or instructions – then translate that into code. If you have an algorithm, you don’t need AI. Because you have an algorithm.
And one characteristic of AI is that you don’t really know how it’s working things out under the hood. It’s generating its own heuristics.
It takes a lot to get me to clear a path for the AI hype train, but he’s only been and gone and done it.
Completely agree with all the comments on AI limitation but from Murphy’s perspective it suggests to me:
– the grift is failing and he is short of cash
– He is relying on inheriting his ex wife’s pension which would be at risk if AI took her job
– The recent election means his chance of Ermine with the SNP has gone and he has burned his bridges with Labour so despite his ideology being in the ascendancy he is personally out in the cold
I do hope this is the case!!
Shirley, the important thing is not how many mistakes AI makes but how many the non-AI human meatware makes? And which one makes less. I don’t think, on it’s past record, my money would be on the meat.
The eBay example that WB gave is a rather good one. What would happen if you offered your title to a flesh & blood eBay employee? eg Do you know what a draw plate is? eBay ML got me a selection straight off.
So an AI that has your total medical history and the entirety of medicine to draw on and it will fail….. man’s an idiot
In contrast, the whole point about medicine is that it takes seriously incomplete information, the vast majority of it being communicated nonverbally, and interprets that based on the intuitive experience of the practitioner.
Lol. AI better than humans at recognising teh gayers:
https://www.theguardian.com/technology/2017/sep/07/new-artificial-intelligence-can-tell-whether-youre-gay-or-straight-from-a-photograph
Labour Government: “We can grow the economy by unleashing the potential of technology and AI to increase productivity”.
Peeps: “So how do we do that then?”
LG : “By unleashing the potential of technology and AI to increase productivity”.
As someone once said “It’s all slogans, buzzwords and meaningless drivel”.
These cunts do not have a clue.
Addolff
Genuine ROFL at the last line – that’s fucking brilliant
Most of these fuckers wouldn’t know an AI if it was wearing a LGBT flag and rogering them up the ass.
And we have at least five more years of this shit.
We know some medics retire in their 50s ‘cos high taxes, or they can, but I wonder if the NHS being an employer who is never happy and boredom are factors too. With AI assistance the job is going to get seriously interesting for doctors staying in the game till their 70s as the referrals will come from lesser consultants who are stumped. Good for recruitment and retention. One can hope.
“ The weightings provided are those that at the moment when the decision has to be taken (3am at the bedside of an unconscious elderly person with multiple co-morbidities) seem best to the doctor using a combination of all the heuristics that years of experience has provided to them, many of the inputs for which they will never have time to record.”
Anyone who’s worked all nighters will know that human beings are not the brightest at that time, add in the Dr might be worried her husband is shagging the Aupair, or a male doctor thinking about how he might get his end away with the good looking new nurse, etc and I think I’ll take my chance with AI.
“Shirley, the important thing is not how many mistakes AI makes but how many the non-AI human meatware makes? And which one makes less. I don’t think, on it’s past record, my money would be on the meat.”
But will people accept it? Its the same argument as with the smart motorway issue – statistically you are supposed to be safer on a SM, but thats not much comfort to the family of the man killed when his car breaks down and gets ploughed into by a lorry because there’s no hard shoulder.
If Granny gets offed by an AI doctor giving her the wrong meds entirely are people going to accept that, on the basis of ‘the greater statistical good’? Or are they going to kick off, and as with the SM case, those who have suffered losses get listened to, because its impossible to listen to those who might have been killed by a human doctor, but weren’t by the AI one.
And as I keep asking, who is liable for the AI doctor’s errors? Maybe the way forward is to give the AI brigade some skin in the game – every patient they kill or seriously injure through their errors gets an automatic £10m, no questions asked. That’ll make them think a bit before saying how wonderful and accurate their medical AI model is.
“good looking new nurse, ”
Oh mate, we do not live in a Carry On film
( actually we do… Screaming )
Nevertheless ‘good looking nurses’ these days are difficult to find, especially ones under 14 stone.
Most of these fuckers wouldn’t know an AI if it was wearing a LGBT flag and rogering them up the ass.
True, but that differs from the other parties (and I’d include Reform), how? If the Conservative plan is (as it has been for the last 14 years) “we’re just the same as Labour, but a bit less enthusiastic about it”, I can’t see that playing well.
We badly need some folk with a clue in government (both the political and executive sides), but there’s precious little sign of them. Just PPEs and African Studies degrees as far as the eye can see.
This precise requirement was the basis for a use case for Watson (the computer that won Jeopardy) as the first natural language processor way back in ancient computer history (2011)
The original has now been sold off because of the last decade+ advancements it sparked off but you can see the latest iteration at https://www.ibm.com/topics/artificial-intelligence-medicine
Comments are closed.