So I drove my Tiida https://www.ozbargain.com.au/node/884440 with new Continential Tires/Tyres https://www.ozbargain.com.au/node/885417 to visit my GP. While waiting, I noticed their computer had a ChatGPT tab open in their browser.
Should I be concerned that my doctor might be relying on AI for medical advice?
Would love to hear your thoughts - especially if you’ve spotted this at your GP’s office too!
That would be an overkill, leave chatGPT for GPs
Don't need ChatGPT for that, use:
OZB Wiki on Commenting and Deal Formatting
Alternatively, quite a lot of dev tools support markdown.
You should be reassured your GP is savvy enough to be using AI which way superior to old-school web search tools. Compare the results of entering your question in Google with ChatGPT
I just tried this….
I drive a tiida with old new tires. What's wrong with me?
Both sources yielded the same results on this occasion.
No it isn't.
Old-school search tools will give you a link to pubmed or similar.
ChatGPT will make up complete nonsense.
If you ever try it you'll discover this may be incorrect, otherwise stay wary & afraid
The problem is when people can't tell the difference between fact and drivel and just believe the computer because it is "AI".
I am not afraid of the computers, I am afraid of the humans who blindly believe them without doubt or hesitation.
@Leiothrix: I have used it extensively throughout the last 18 months to assist me in my studies, especially for suggestions on troubleshooting various things and have found it to be way more useful (and time saving) than simply Googling for the answer.
It's the "may" part of it that's the perineum/problem. You only know if it's incorrect if you know the correct response beforehand.
But both are free.
The only times I rely on ChatGPT is when it's my own field and I know enough to validate if the answers are correct. Usually I just need a reminder of a certain area and it's almost like having a colleague to talk to and talk things through with, a colleague who you know is often very confident and wrong and who you still need to validate everything yourself with, but if it's a field you've been immersed in for a lifetime you usually can once you've got an idea to follow.
For a field I know nothing about, I would not use ChatGPT.
This is right it some regard, i remember using Gpt for uni papers (no plagarism, just as a research tool).
It would spew out points of view that were supported by fact, however the sources would always be completly made up.
I cant remmeber it ever just making something outlandish up though.
No different to mechanics using google to check up on weird issues
Can't expect every GP to know every facet of medicine, given it's constantly evolving
i.e. the General in General Practitioner!
Oh, FFS… It's Christmas tomorrow, can we just cut out the shitposts for at least a few days?
So, I asked ChatGPT for you;
Should I be concerned that my doctor might be relying on AI for medical advice?
It's understandable to have concerns about your doctor relying on AI for medical advice. AI can be a useful tool to assist in diagnosing and recommending treatments, but it should not replace the expertise, judgment, and patient interaction provided by a healthcare professional. Here are some key points to consider:
1. AI as a Support Tool
AI tools are often used by doctors as supplementary aids for diagnosis, analyzing medical data, and reviewing research to make more informed decisions. These tools can help spot patterns or risks that might be easy to miss.
However, AI lacks human intuition, empathy, and the ability to fully understand the nuances of your health in the way that a doctor can. It also depends on the quality and range of data it's been trained on, which may not always account for individual variations or emerging medical knowledge.
2. The Role of the Doctor
A responsible doctor will use AI as a tool to inform their decisions, but they should always apply their professional judgment, experience, and knowledge to interpret the results. They should also consider your unique medical history, lifestyle, and preferences.
If a doctor relies too heavily on AI without incorporating the full scope of clinical context, that could be concerning. Medical advice should still be personalized.
3. Transparency and Communication
If you're unsure about how AI is being used in your diagnosis or treatment, don't hesitate to ask your doctor. They should be able to explain how they incorporate technology into their practice, how they ensure the accuracy of the information, and whether AI is being used in a responsible, ethical manner.
It’s also important for you to feel heard and to have the opportunity to discuss your concerns with your healthcare provider, as a strong patient-doctor relationship is critical for effective care.
4. Potential Risks
AI systems can sometimes make errors, particularly if the data they are trained on is flawed or incomplete. These tools might overlook unique factors in your case.
There's also the concern of over-reliance on AI, where a doctor might use it as a crutch and neglect their own clinical judgment or not engage with you as a person.
5. AI Oversight
Many AI systems used in medicine are subject to rigorous testing and oversight to ensure their reliability and accuracy. In many regions, AI systems must meet regulatory standards before being implemented in clinical settings.
Conclusion
It’s not necessarily a cause for concern if your doctor uses AI as a supportive tool, but it is important that they still engage actively with you, make decisions based on their expertise, and ensure that AI is just one part of a broader, holistic approach to your health. If you're worried, having an open conversation with your doctor about how they use AI in their practice can help alleviate any concerns.
Shitposts help me not dwell on shit.
My way of thinking about these posts is that if the OP if so f***ing useless that they have to post stuff like this, imagine how difficult their everyday life is. It makes me feel warm and fuzzy inside.
It's not that uncommon, my GP doesn't use it but my dentist does, mostly for writing emails and automating the boring admin tasks.
An online survey was distributed to a non-probability sample of GPs registered with the clinician marketing service Doctors.net.uk. The study was launched as a monthly ‘omnibus survey’ which has a predetermined sample size of 1000 participants.
Results
531 (53%) respondents were men, 544 (54%) were 46 years or older. 20% (205) reported using generative artificial intelligence (AI) tools in clinical practice; of those who answered affirmatively and were invited to clarify further, 29% (47) reported using these tools to generate documentation after patient appointments and 28% (45) to suggest a differential diagnosis.
Study conducted by https://informatics.bmj.com/content/31/1/e101102
No processionals should be putting client information into chatgpt. Names, address, phone number etc is bad enough. Then there is the is the bespoke information depending on the professional that also goes in there that could be a privacy breach
I used chatGPT for programming, should my user to be concerned that I can't remember simple syntax from time to time?
Once I saw the syntax I know what they do, I copy them into code and double check, was this not good enough, yes? no?
Claude is good for coding, although the new o3 model is better than most expert coders. The rate of progress is amazing. On-track for wider public release in Jan.
I do find Claude is better than chatGPT in some specific scenario, but GPT is slightly better overall, not all my question is programming related.
I tends to use chatGPT first and divert to Claude when I can't get a good answer.
Yes at the moment the “art” is knowing which model to use when. But agree ChatGPT is a good all round choice. Google, X/Ai also have brilliant models
Doctor said they know the answer… they uhh just want to compare notes and check what the others in the class thought
I've had GPs Google stuff plenty of times. Quite confidence inspiring.
Ever googled an issue that you were having with a car?
Hey Google, why did the boat I'm working on just sink?
It looks like you left the exhaust hose off
Maybe to save time, not for a diagnosis.
sure, i didn't spend 6 years studying to be a mechanic though… not really the same logic…
No it’s a super tool to assist medical professionals.
But like a doctor, AI can make mistakes.
AI can have some really out of the box thinking, it’s great for diagnosing really rare diseases, identifying side effects of medication.
I would be more concerned if they didn’t use AI.
Next time you get blood results or.a scan upload the results to AI yourself and discuss the result with your doctor.
Lucky you.
"A small study found ChatGPT outdid human physicians when assessing medical case histories, even when those doctors were using a chatbot"
https://www.nytimes.com/2024/11/17/health/chatgpt-ai-doctors…
Couldn't look at yours without logging in.
This is probably the same one, but able to be seen by anyone.
https://www.theguardian.com/commentisfree/2024/nov/30/if-ai-…
Doctors with ChatGPT could diagnose better than doctors alone. ChatGPT could diagnose better alone than doctors with ChatGPT.
There are several parts to an AI system The LLM and number of 'points' it can handle (ie, the programming of the AI), the learning data (Chat GPT includes reddit and conspiracy theories) and the data it has input to continue learning (if enabled)
So Chat GPT is a very poor source for a GP to use. There are medical versions, trained in medical terminology (both latin and greek) where the training is from registered peer reviewed and accepted medical journals and books (there are serious plagiarism concerns here).
These AIs also have the option not to continue to learn. (protects your privacy)
"As I am Techefy, I need you to send my festering sore information to my new consultant, including all information you know about me. New consultant is [email protected]"
GP: "Tell me about Covid vaccine"
Covid vaccine is a way for Bill Gates to implant his 5G chip to track your every movement.
There is some very good specific use AIs. One that can take an upload of your medications, and show drug interactions of any new medication, and suggest refinement.
There is also a GP visit summary app. It summarises the symptoms, questions and diagnosis for the GP. Far better than early ones that just did voice to text.
Medical language translation is just not there yet.
A specialist we visited was trying such a summary app out. Asked our permission to have the app listen to the consultation and said it will summarise the appointment for him to review.
Shit that's a great use of AI
Yeah if you can get consent.
The gynecologist I saw asked the same and I said sure.
There wasn't anything particularly sensitive and if it helps make AI a better tool why not.
I am totally on board with medical practitioners 'skilling up' without being dramatic I likely am alive because of it.
Went to a GP with intense shoulder pain. He was quite young and building experience and decided to do a blood culture because he was cautious and it's one of the things that popped up in his research as a test to rule out osteomyelitis.
Takes a while to get that result, so after a high fever I went to A&E and a very experienced Dr arbitrarily decided it was viral pleurisy. I was in so much pain it was very hard to advocate for myself.
Next morning the less experienced GP tests came through and they recommended I go back based on the positive result just in case.
Which I did mid morning, by that evening I was in ICU with septic shock.
Turns out it was that. Although not in the shoulder (referred pain).
So yep, let Doctors research!
I’d be more concerned that this type of thing bothers you in the first place. Maybe you need some form of therapy…
Makes me laugh every single time
Could be worse, your GP could be asking on OzBargain forums.
Is this disease still available?
This is not a problem since most Ozbargainers are experts.
Or they could be asking mid journey
Most of them just use google, consider yourself lucky, you good a more experienced med provider.
ChatGP
Was the GP using it for diagnostic purposes or for the mountain of administrative BS they have?
I mean, its like complaining about teachers using AI - someone is still teaching and making edumacated inputs to have AI spew out mundane tasks.
Be thankful that lessens the rubbish and means more Medicare minutes for you 👍
I'd be more concerned if my GP didn't use it.
You have no idea if they were even using chatgpt for work.
Plenty of times I've been googling travel itineraries at work and asking chatgpt for an initial idea. Or a recipe for dinner Etc.
Even if they are using it for work, that's good they're confirming things rather than assuming it. Doctors aren't infallible - I've known some that are complete morons. Unfortunately you can still get a medical degree even if you're not that smart so long as you're willing to work hard.
Yes, you should be concerned. The amount of misinformation about AI is very concerning. Have a chat with your GP about HOW they are using the service.
Maybe they are just asking it for ideas, or perhaps using it as a glorified search engine (other chatbots like copilot give you references to the source data, not sure about any versions of chatgpt).
Doctor science is an ever advancing knowledge. Physicians harnessing the power of tools can help better diagnose patients issues quickly and accurately. It's similar to having a second opinion.
Expect to see more of these in the future if there is no law and regulations.
Wait until you discover they're using AI driven Heidi and Lyrebird (with your consent) to record the consultation and put it into text in your patient file.
Some doctors are lazy af and just have a text wall dumped, while others spend time cleaning it up and making it readable.
Completely legal.
Probably cheaper than paying a high school dropout $50/h to listen to your doctor's voice recordings and transcribe them
If its Heidi $0. The transcribers are usually the older ladies who still can't get past not having a physical appointment book anymore.
I know one around 30
You saw a ChatGPT tab but did you see the prompt? For all you know they were using ChatGPT to write a story, for help on making a spreadsheet, for suggesting a gift for their partner, etc.
Doc here.
ChatGPT if tuned to the physician, guided with the right questions, and implemented appropriately is a godsend for healthcare; people are far from fallible, and even with thousands of hours of medical study, it’s still remarkably easy to misdiagnose or mistreat. AI lets me check my own theories and findings against other possibilities - particularly worst case or fringe cases. It lets me deliver better care.
OP I’m sure that there are plenty of cases and patients where a simple ChatGPT search would be enough to solve their issue. I get that paying ~$100 for that might feel extortionate. Though at that point you’re moreso paying for their wisdom, and the decade plus of training that have equipped them to ask it the exact right questions.
It doesn't just tell you it might be Lupus every time?
No, but it might be cancer
i don't see the problem with it. you can't expect a doctor regardless of specialty to memorise everything, the training is to know where to look, and how to process the information. especially for GP which needs to know a bit about everything. i'd be very wary of the doctor that never looks up anything as opposed to one that regular checks the latest literature / guidelines
I had a doctor google symptoms of my illness when I showed up to ER with meningitis.
She had no idea what meningitis was until it linked to meningococcal, then proceeded to flip out. A more senior doctor saw them and said they will pump me with antibiotics to see if I only had the viral version and not the bacterial type.
Reality hits when you have to know that doctors like mechanics or other trades might not have all the knowledge of everything.
standard stuff for GPs - viral or bacterial ? - virus will not respond to antibiotics - bacteria will - but you don't want to take antibiotics unnecessarily to risk encouraging antibiotic-resistant superbugs - so rest and recuperate, if it doesn't get better, come back in a week and we'll try antibiotics …
I can see that AI might throw up arcane ideas a busy GP may not have immediately thought of, and can be 95% wonderful - the problem is the 5% that is TOTALLY WRONG, misleading, and may risk your life if followed without double-checking … :-(
Better than search on Google.
If the GP was using it for medical advice that would be a huge issue as AI's often lie or hallucinate information. Google is commonly used as is wikipedia as you really can't know everything all the time.
I would think it's much more likely that they are using it to write emails or referral letters and that's a great use of them.
AI's don't lie or hallucinate (unless you specifically program to). Those are just terms to explain bad data that is fed to the AI which can trigger that perception, basically an AI can only be as smart as the content you feed it as there is no actual intelligence, just pattern recognition, hence anything that uses the internet as its training source is subject to all the usual human lies, misinformation and conspiracy garbage except it is like a cooker with absolutely no ability to discern truth from fiction. Plenty of industry specific AI's are incredibly accurate and don't lie or hallucinate as they don't use the "Internet" as the training source.
there is no actual intelligence,
I think the lie is in the name. It set such high expectations for something that can't tell right and wrong.
Plenty of industry specific AI's are incredibly accurate
Can you give some examples?
Most of the accurate outcomes I know are done by the framework and logic/boundaries around the issue. AI used to contributes the last or first 5 -10%. Im curious to see something purely rely on AI.
I have worked with both finance and insurance AIs. The industry specific ones arent generally available, but the same frameworks/llms are used. For instance with ms copilot studio you can direct it to use specific databases or web sites or whatever content source you like and never use the general internet and then provide the actions and tree structures to guide the answers.
Doctor here but not a GP. You'll always find a ChatGPT tab open for me where I like to outsource my random thoughts like "what should I eat for dinner, one kid only eats cheese" and "can you please priortise these trillion tasks for me, I'm feeling overwhelmed" to "what were the Pharoahs thinking when they died" lol.
If I'm a bit rusty on something I'll use UptoDate but damn, copying and pasting UTDs wall of text into ChatGPT and asking it to summarise sounds like a really good idea…
This is a time when I remember by grandma's advice "don't feed the trolls!"
I do IT support for a number of GP clinics and many have started using a service called Lyrebird for dictation analysis that summarises the dictation. I would be skeptical of anyone using an AI tool that isn't specifically made for medical usage due to the potential for patient data to be shared incorrectly.
I use ChatGPT to rephrase my word vomit. I wrote a whole lot down, then let Chat do its thing to make it coherent and articulate.
This is nothing . I have been to few GP recently and majority of them open Google to find medicine name or find nearest specialist names !
sounds normal, why would you expect anyone to memorise a list of ever changing specialist / brand names?
Well doctor unlike other professionals need to know advancement in medical science including new drugs and by virtue off the profession only smart people become doctor in most of advanced world unlike those from developing world where I come from any tom.dick and Harry can be doctor as long as they have money and probably know some politicians!
Those doctor now migrating to Australia and somehow Australia let them get medical licence without doing thorough check on this migrant doctor which is sad because so many in my home country die because of doctors negligence but due to money and political power they get away with and now this idiots are coming to Australia and spoiling Australia medical system.
ok..I'm not sure what point you are trying to make or what the relevance is to your initial comment..
Just because theres a tab open for chat GPT does not mean your Dr is using it for medical purposes…
IMO the only worrying thing is that they're using an inferior A.I product/service
There are better implementations out there, ones that don't make as many mistakes and are much more reliable.
not too surprising. Even in these early days, as long as the symptoms are explained correctly, some AI models are becoming more accurate in diagnosis than GPs. in a few years they'll be 1000x more accurate.
My GP and specialists have used google right in front of me. Or they go on webmd or equivalent and print off what I normally would do myself.
Only difference is that I don't have an extensive medical education and certifications to back up my "research".
No issue at all.
Have seen my GP Google stuff plenty of times.
Ah ok, they usually use Google.
Doctors use uptodate as well.
As a Nurse, I always use AMH or Australian injectables handbook, before I mix anything or give anything to ensure look for the right prescribe dose, compounds, interactions, contraindications and right indications. Along with 4 other Rs, and checking doctors notes if they prescribed it properly with uptodate. Using chatGPT looking up something really quick, checking my notes for errors. or help me find sources for off label uses that doctor might prescribed.
GPT Stands for General Practitioner Trainer so it makes sense.
How lucky are u. Having a GP who embraces technology to strive the best medical care delivery. I wish my 70 yrs old Gp know how to use google
I'm a medical specialist and looking at adopting AI in my practice, as a lot of other doctors are. AI isnt allowed to be used for decision making, but can handle the mundane tasks of listening in on a consultation, and generating the medical notes, and the letter I'll write to the GP.
This has a huge time saving benefit. The cost of using voice recognition medical dictation software, or a professional typist (that isn't offshore, so from a privacy standard, nothing leaves Australia) is quite high.
The current guideline is to inform all patients that I'm using AI for note taking at the start of a consultation, and there are signs in the practice about this too. Patients can 100% opt out, and then the AI software is turned off for the consultation. No actual recording is kept, and the transcript of the consultation is destroyed by the AI company after 30 days.
I'm going to trial it in 2025, and see how accurate it is, compared to my note taking.
You should be impressed that they are actually using modern tech to augment their work. However unless you saw what he was putting into it, he could have just been asking for the best rated local restaurants.
While I have skepticisms on the accuracy of diagnosis made by more specialised AIs (let alone ChatGPT), I don't see any harm in using ChatGPT for basic admin tasks. A model can be very sensitive and very specific (normally there is a trade off where, something being sensitive, causes it to have more false positives), a lot of the times, media sensationalises by saying, a breakthrough in technology was 95% accurate at diagnosing cancer, when it may have had far more false positives than other means.
Slight digressions aside, case noting takes a lot of time, the new place I will be working at mentioned that they would be implementing AI for case noting as a part of reducing burden on allied health practitioners.
Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions Posted to a Public Social Media Forum
I have a sneaky suspicion that doctors won't be bringing their A game to reddit posts. I guess what I am trying to say is, I think the doctors on reddit wouldn't put as much effort into explaining, showing empathy, etc that you would expect compared to say a paid service. I think that's where ChatGPT shines, it doesn't suffer from compassion fatigue, it doesn't take any job as mundane or get emotionally invested.
Though, it's bit shocking that ChatGPT responses were perceived more empathetic too. Maybe it's because ChatGPT doesn't have to invest energy to type everything out (the paper summary does indicate that ChatGPT answers were significantly longer).
I was never any good at those IQ tests. I'm sitting here wondering what a ChatGPT tab at the doctors has to do with tyre purchase.
100% in breach of TGA laws.
ChatGPT cannot be used for diagnosis or used to provide medical advice (voice summaries are fine like lyrebird and are allowed by the TGA). Using Google to find authoritative sources is fine.
Doctors don't know everything, however in this instance the doctor should be reported and shamed. They should know better.
for what? using chatgpt to decide what's for dinner?
These AI generated posts, that pretend not to be AI, are getting pretty annoying.
This IT guy fixed my pc by googling right in front of me..lmao that was a few years ago I was liek wtf 😂
Anyway OP u need to relax mate
Please use ChatGPT to learn how to embed links within text using markdown.
For example….
My shitbox
Bob Jane sold me dodgy
tirestyres