So glad you’re okay!! This does, in my opinion, say far more about the state of general/mainstream medical knowledge (and possibly the state of our insurance system) than it does about LLMs. If you ever notice a neurological symptom: go get checked out. Period. A neuro symptom is not like a sore muscle or a tummy ache. It’s almost always a sign of something serious. I would’ve immediately gone to an urgent care or ER when I noticed the droopy eye.
What an incredible story. A few months back I remember a post on Twitter going viral about a guy who made it to the ER just in time after ChatGPT flagged his testicular torsion. Did you use the Thinking model for this, or was it just the base one?
I'm a harsh critic of advanced AI in general and LLMs in particular. I think there are nearly a dozen distinct kinds of existential threats that they pose to society. Given that, this anecdote is good for keeping my zealotry in check. I'm glad that you're doing well!
Thank you for sharing this amazing story. I’m glad it’s had a happy ending for you, Ben!
You inspired me to do my own research, which has been fascinating and has actually reduced a fair amount of stress for me.
I’ve lived with an arrhythmia for years and have been at an impasse in recent months. I’ve known intuitively that the nervous system is impacting the episodes of AFib, but the cardiac specialists tend to be very narrowly focused and haven’t always been willing to try to understand the complexities of my situation.
Of course it makes sense that ChatGPT would have a take that makes the connections I’ve sensed for years since that tends to be what AI does. I will definitely make my own decisions and am not blindly following anything, but have found the various suggestions, including treatment options and language to communicate my needs respectfully and clearly, to be extraordinary.
What’s really wild is that I have actually felt more seen, heard, and the complexity of my situation understood by AI than some of my doctors. Wow. Registering that is a trip.
Thank you for sharing your story, Ben! I wonder if others have also been inspired to do their own research.
My husband had an aortic dissection around the year 2000. He was a scientific researcher and was accompanied to the ER by a colleague who had trained as a Neurologist (who took his symptoms seriously). Despite this, the ER doctor sent him home with information about migraine headaches! I worked in medicine too, as a Nurse Practitioner, and was appalled (at the time the standard of care was that you needed a history of several similar episodes to diagnose migraine, and Jon didn’t have this history. We tried to get a second opinion, but it was a drop in appointment with a provider who had never met my husband, and she backed up the ER doctor. I had to put up a fuss (not my nature, I had to become the problem wife) with the health system (HMO) in order to get him a carotid ultrasound, at which point the “system” began to take his affliction seriously. 6 months of anticoagulation while the artery healed, thankfully no long term effects.
AI does medical diagnosis better than humans, according to the latest experiments. Something like MD's get it right 75% of the time and The MD LLM gets it right 90% of the time.
Happy for you, my friend! Many may disagree with me, but I find the quality of health results I get with Dr GPT to be much higher than those I get from a regular doctor (who rarely does nothing more than want to write a prescription… ugh). I don’t want a pill, I want an answer and a solution. Best success to you!
Great comment and something I have been suspecting for a while. Non-surgical and non-hospital doctors are likely going to turn in to more like coaches who may try to help people prioritize their needs and steer them towards overall health improvement. The hospital is happy to take care to people in actual danger and those who think they are in danger for a very high fee. ChatGPT is will help you avoid the doctor for very minor things but who will do the in between work like managing blood pressure medication or diabetes? I think the next generation of nonsurgical doctors would actually like it if patients asked AI what the best medication would be and how to titrate it, print the answer out and give it to the doctor. If it makes sense to them, they will happily help you get the medication and tests done to test the minor condition safely. The current group of doctors like myself have spent years compiling data in our brains that Dr GPT will give to you for free. It is actually better than us because it doesn’t have off days. My question is: In the future, who will actually write the prescription and take responsibility when something unexpected occurs? Pretty much the same question that applies to self driving vehicles.
When I imagine going to the ER and saying ChatGPT sent me, or telling them "hi I think I might have this very specific medical problem, no I'm not a doctor or anything" I feel almost prohibitively embarrassed in advance, based on reactions I've gotten from doctors when bringing up hypotheses based on independent reading. Do you happen to remember what specific words you said to the ER folks when you got there? How did they respond? Did you mention ChatGPT? How'd you explain your hypothesis?
I remember going to the doctor in Thailand for an infection after dr.chatgpt told me to. The doctor said there were two possible options, but he preferred to rely on the chat's answer
I was also sent to the ER by chat GPT due to a weird headache! In my case, no serious issues (but chat’s recommendation was in line with clinical guidelines)
So glad you’re okay!! This does, in my opinion, say far more about the state of general/mainstream medical knowledge (and possibly the state of our insurance system) than it does about LLMs. If you ever notice a neurological symptom: go get checked out. Period. A neuro symptom is not like a sore muscle or a tummy ache. It’s almost always a sign of something serious. I would’ve immediately gone to an urgent care or ER when I noticed the droopy eye.
What an incredible story. A few months back I remember a post on Twitter going viral about a guy who made it to the ER just in time after ChatGPT flagged his testicular torsion. Did you use the Thinking model for this, or was it just the base one?
Oh, good question! It was GPT 5, Thinking. And thanks!
I'm a harsh critic of advanced AI in general and LLMs in particular. I think there are nearly a dozen distinct kinds of existential threats that they pose to society. Given that, this anecdote is good for keeping my zealotry in check. I'm glad that you're doing well!
My goodness! So happy you went to the ER and avoided a possible stroke! Hope you have no further issues. 🤞
Amazing. So glad you're alright!
Thanks! Me too :)
Thank you for sharing this amazing story. I’m glad it’s had a happy ending for you, Ben!
You inspired me to do my own research, which has been fascinating and has actually reduced a fair amount of stress for me.
I’ve lived with an arrhythmia for years and have been at an impasse in recent months. I’ve known intuitively that the nervous system is impacting the episodes of AFib, but the cardiac specialists tend to be very narrowly focused and haven’t always been willing to try to understand the complexities of my situation.
Of course it makes sense that ChatGPT would have a take that makes the connections I’ve sensed for years since that tends to be what AI does. I will definitely make my own decisions and am not blindly following anything, but have found the various suggestions, including treatment options and language to communicate my needs respectfully and clearly, to be extraordinary.
What’s really wild is that I have actually felt more seen, heard, and the complexity of my situation understood by AI than some of my doctors. Wow. Registering that is a trip.
Thank you for sharing your story, Ben! I wonder if others have also been inspired to do their own research.
Sounds like a movie…..😊
Glad you’re okay! It’d be interesting to see how you wrote the prompts in these situations
My husband had an aortic dissection around the year 2000. He was a scientific researcher and was accompanied to the ER by a colleague who had trained as a Neurologist (who took his symptoms seriously). Despite this, the ER doctor sent him home with information about migraine headaches! I worked in medicine too, as a Nurse Practitioner, and was appalled (at the time the standard of care was that you needed a history of several similar episodes to diagnose migraine, and Jon didn’t have this history. We tried to get a second opinion, but it was a drop in appointment with a provider who had never met my husband, and she backed up the ER doctor. I had to put up a fuss (not my nature, I had to become the problem wife) with the health system (HMO) in order to get him a carotid ultrasound, at which point the “system” began to take his affliction seriously. 6 months of anticoagulation while the artery healed, thankfully no long term effects.
Thank goodness you used Chat GPT.
AI does medical diagnosis better than humans, according to the latest experiments. Something like MD's get it right 75% of the time and The MD LLM gets it right 90% of the time.
Not a super useful metric I would say… unless you know the false positive rates as well?
Happy for you, my friend! Many may disagree with me, but I find the quality of health results I get with Dr GPT to be much higher than those I get from a regular doctor (who rarely does nothing more than want to write a prescription… ugh). I don’t want a pill, I want an answer and a solution. Best success to you!
Great comment and something I have been suspecting for a while. Non-surgical and non-hospital doctors are likely going to turn in to more like coaches who may try to help people prioritize their needs and steer them towards overall health improvement. The hospital is happy to take care to people in actual danger and those who think they are in danger for a very high fee. ChatGPT is will help you avoid the doctor for very minor things but who will do the in between work like managing blood pressure medication or diabetes? I think the next generation of nonsurgical doctors would actually like it if patients asked AI what the best medication would be and how to titrate it, print the answer out and give it to the doctor. If it makes sense to them, they will happily help you get the medication and tests done to test the minor condition safely. The current group of doctors like myself have spent years compiling data in our brains that Dr GPT will give to you for free. It is actually better than us because it doesn’t have off days. My question is: In the future, who will actually write the prescription and take responsibility when something unexpected occurs? Pretty much the same question that applies to self driving vehicles.
Great comment. Who will be responsible for AI opinions or errors?
And who will be happy to pay for all the likely slew of "unnecessary" (excessive, defensive) investigations that AI triggers?
Thank you for this writeup!
When I imagine going to the ER and saying ChatGPT sent me, or telling them "hi I think I might have this very specific medical problem, no I'm not a doctor or anything" I feel almost prohibitively embarrassed in advance, based on reactions I've gotten from doctors when bringing up hypotheses based on independent reading. Do you happen to remember what specific words you said to the ER folks when you got there? How did they respond? Did you mention ChatGPT? How'd you explain your hypothesis?
Amazing story! Your description of the medical care you received was moving! So good to hear!
AI reminds me of a scalpel. In the hands of a surgeon it can save a life. In the hands of murderer it can take a life.
This is the good side AI at work
Good to hear you are okay!
I remember going to the doctor in Thailand for an infection after dr.chatgpt told me to. The doctor said there were two possible options, but he preferred to rely on the chat's answer
I was also sent to the ER by chat GPT due to a weird headache! In my case, no serious issues (but chat’s recommendation was in line with clinical guidelines)