4 ms·

ChatGPT medical advice brought man 'to brink of death', lawsuit alleges

23 points by dijksterhuis 3mo ago

lardosaurusrex 3mo ago
The amount of times these LLMs randomly refuse to answer basic questions because "omg safety" only to turn around and do this f￭￭￭ing nonsense is obscene. I'm not blaming the LLM because it's just a pile of math glued together with hubris and anxiety but why is its default response not "I am not a medical professional and you should see a doctor right away if you have any concerns"?
- fhn 3mo ago
  because they get sue otherwise?
sheepscreek 3mo ago
I really hope the lawsuit does not end in OpenAI tightening the “safety” and “security” of their models into oblivion. I hope they understand that a lobotomized model is as good as my Casio watch. I want my models to give it to me straight without patronizing me. I’m an adult and I take responsibility for my own actions.
- lardosaurusrex 3mo ago
  That's the issue here tho. You do not own the model; they do. Ergo it is their advice and so they can be held accountable fo it.
  - fhn 3mo ago
    so if you give your friend medical advice and they are injured, can they sue you?
    - dnemmers 3mo ago
      I'm sure they could try.
    - lardosaurusrex 3mo ago
      [flagged]
      - SR2Z 3mo ago
        No he's got a point, if your friend and OpenAI both preface their statements with "I'm not a doctor but..." then there is no reason for it to be ok when one does this and not the other. We're several years into having LLMs be publically available. If people still don't understand their limitations it's not the models' fault. This situation in my mind is kind of similar to a random person googling symptoms and then brainlessly reading the first website they find. I think that argument is pretty persuasive, actually, because I use these things all the time and am well aware they're prone to bullshit. I cannot fathom someone with unstable blood pressure and other unexplained symptoms not going to see an actual doctor JUST BECAUSE ChatGPT told them they were fine, and it makes even less sense that anyone would listen to a soulless machine talking about "God's plan." FFS, when you have serious medical symptoms, JUST GO TO THE ER. They get plenty of people convinced they swallowed their earbuds while sleeping, someone with fluctuating blood pressure is so much more serious. If the argument you're making is "some people are too stupid to be trusted with an LLM" then you should say that, but otherwise this is crazy.
        dijksterhuis 3mo ago
        this situation says more about rampant/unchecked/unregulated businesses and the american legal system/beliefs around regulation than anything. mcdonald’s coffee “warning: contents may be hot” comes to mind as a previous example. fast food businesses were competing to serve the hottest coffee and one of them ended up scalding a woman so badly they were forced to put warnings on their coffee cups. replace coffee being hot with LLMs. replace scalding with stopping someone seeking medical assistance. but it’s okay because #murica #trustUsBro #profitsBeforePeople /s
        SR2Z 3mo ago
        Nobody would expect their coffee to be almost boiling, everybody should expect an LLM to give slop answers from time to time. Even Steve Jobs died early because he took vibes-based advice instead of just listening to doctors. You can snark about capitalism all you want, but at some point people are in charge of their own health.
  - sheepscreek 3mo ago
    I don’t think this is true if a disclaimer is provided. Most investment advice videos on YouTube make this very clear, but it doesn’t stop them from sharing advice. I suppose there is some precedent for this in the American justice system.
harvey9 3mo ago
All the chat bots have a very discreet disclaimer at the bottom of the page. I want to say it should be bigger text and more strongly worded but like most people I would just mentally tune it out regardless.
- Arnt 3mo ago
  The LLM I use is fairly quick to retreat into extra disclaimers. "Not sure about this, this is from memory and you ought to check", etc. Don't they all do that? Is it just because of my prompt (which does say to push back if I appear to have bad ideas)? It seems possible that people tune out disclaimers and caveats in the middle of the answer, too.
arm32 3mo ago
Why are people even allowed to dress up ChatGPT... an autoregressive transformer... as a faith-based holistic doctor? Anybody could see this going wrong from a hundred miles away.
- throw310822 3mo ago
  A few days ago a family member insisted to be brought to the ER because Claude told him. Result: he was immediately hospitalised and received surgery the following day to remove a tumour that was compressing his spine. Claude was very clear that he had to act as quickly as possible to lower the chances of permanent damage.
heaney-555 3mo ago
This was with GPT-4o (ancient now by AI standards). The reason OpenAI refuses to bring it back, despite all the demand, is that when the conversation gets long enough it starts doing weird things like this. Subsequent models have almost entirely eliminated this kind of thing.
- siva7 3mo ago
  Even GPT-5.4 is ancient compared to intelligence of 5.6 although those are only 4 months apart. We're talking about at least twice the capabilities and that's scary to achieve in such a short timeframe.
fhn 3mo ago
this is number one bullshit. You ask a medical professional for medical advice. Getting medical advice from AI or anywhere else is your own fault.
sharts 3mo ago
There are no bounds to human stupidity
ButlerianJihad 3mo ago
Ironically, the given name "Claude" means "crippled" or "lame". Students of the Roman Empire may recall the disabled Emperor Claudius.
elisbce 3mo ago
The comments here are ridiculous. Do you know how much ChatGPT has helped families, esp. those without access to the top medical doctors and resources around the world? My own experience with it so far have been nothing but positive. It beat all the doctors in various hospitals that we have consulted so far on breadth and depth and the best thing is you can keep asking follow-up questions on anything and it gives you all the details you need without being rushed. The fact that it is not 100% reliable shouldn't be basis for restrictions. Even top trained doctors are far from 100% reliable either.

23 points by dijksterhuis 3mo ago

The amount of times these LLMs randomly refuse to answer basic questions because "omg safety" only to turn around and do this f￭￭￭ing nonsense is obscene. I'm not blaming the LLM because it's just a pile of math glued together with hubris and anxiety but why is its default response not "I am not a medical professional and you should see a doctor right away if you have any concerns"?

fhn 3mo ago
because they get sue otherwise?

sheepscreek 3mo ago

I really hope the lawsuit does not end in OpenAI tightening the “safety” and “security” of their models into oblivion. I hope they understand that a lobotomized model is as good as my Casio watch. I want my models to give it to me straight without patronizing me. I’m an adult and I take responsibility for my own actions.

lardosaurusrex 3mo ago
That's the issue here tho. You do not own the model; they do. Ergo it is their advice and so they can be held accountable fo it.
- fhn 3mo ago
  so if you give your friend medical advice and they are injured, can they sue you?
  - dnemmers 3mo ago
    I'm sure they could try.
  - lardosaurusrex 3mo ago
    [flagged]
    - SR2Z 3mo ago
      No he's got a point, if your friend and OpenAI both preface their statements with "I'm not a doctor but..." then there is no reason for it to be ok when one does this and not the other. We're several years into having LLMs be publically available. If people still don't understand their limitations it's not the models' fault. This situation in my mind is kind of similar to a random person googling symptoms and then brainlessly reading the first website they find. I think that argument is pretty persuasive, actually, because I use these things all the time and am well aware they're prone to bullshit. I cannot fathom someone with unstable blood pressure and other unexplained symptoms not going to see an actual doctor JUST BECAUSE ChatGPT told them they were fine, and it makes even less sense that anyone would listen to a soulless machine talking about "God's plan." FFS, when you have serious medical symptoms, JUST GO TO THE ER. They get plenty of people convinced they swallowed their earbuds while sleeping, someone with fluctuating blood pressure is so much more serious. If the argument you're making is "some people are too stupid to be trusted with an LLM" then you should say that, but otherwise this is crazy.
      - dijksterhuis 3mo ago
        this situation says more about rampant/unchecked/unregulated businesses and the american legal system/beliefs around regulation than anything. mcdonald’s coffee “warning: contents may be hot” comes to mind as a previous example. fast food businesses were competing to serve the hottest coffee and one of them ended up scalding a woman so badly they were forced to put warnings on their coffee cups. replace coffee being hot with LLMs. replace scalding with stopping someone seeking medical assistance. but it’s okay because #murica #trustUsBro #profitsBeforePeople /s
        SR2Z 3mo ago
        Nobody would expect their coffee to be almost boiling, everybody should expect an LLM to give slop answers from time to time. Even Steve Jobs died early because he took vibes-based advice instead of just listening to doctors. You can snark about capitalism all you want, but at some point people are in charge of their own health.
- sheepscreek 3mo ago
  I don’t think this is true if a disclaimer is provided. Most investment advice videos on YouTube make this very clear, but it doesn’t stop them from sharing advice. I suppose there is some precedent for this in the American justice system.

harvey9 3mo ago

All the chat bots have a very discreet disclaimer at the bottom of the page. I want to say it should be bigger text and more strongly worded but like most people I would just mentally tune it out regardless.

Arnt 3mo ago
The LLM I use is fairly quick to retreat into extra disclaimers. "Not sure about this, this is from memory and you ought to check", etc. Don't they all do that? Is it just because of my prompt (which does say to push back if I appear to have bad ideas)? It seems possible that people tune out disclaimers and caveats in the middle of the answer, too.

arm32 3mo ago

Why are people even allowed to dress up ChatGPT... an autoregressive transformer... as a faith-based holistic doctor? Anybody could see this going wrong from a hundred miles away.

throw310822 3mo ago
A few days ago a family member insisted to be brought to the ER because Claude told him. Result: he was immediately hospitalised and received surgery the following day to remove a tumour that was compressing his spine. Claude was very clear that he had to act as quickly as possible to lower the chances of permanent damage.

heaney-555 3mo ago

This was with GPT-4o (ancient now by AI standards). The reason OpenAI refuses to bring it back, despite all the demand, is that when the conversation gets long enough it starts doing weird things like this. Subsequent models have almost entirely eliminated this kind of thing.

siva7 3mo ago
Even GPT-5.4 is ancient compared to intelligence of 5.6 although those are only 4 months apart. We're talking about at least twice the capabilities and that's scary to achieve in such a short timeframe.

fhn 3mo ago

this is number one bullshit. You ask a medical professional for medical advice. Getting medical advice from AI or anywhere else is your own fault.

sharts 3mo ago

There are no bounds to human stupidity

ButlerianJihad 3mo ago

Ironically, the given name "Claude" means "crippled" or "lame". Students of the Roman Empire may recall the disabled Emperor Claudius.

elisbce 3mo ago

The comments here are ridiculous. Do you know how much ChatGPT has helped families, esp. those without access to the top medical doctors and resources around the world? My own experience with it so far have been nothing but positive. It beat all the doctors in various hospitals that we have consulted so far on breadth and depth and the best thing is you can keep asking follow-up questions on anything and it gives you all the details you need without being rushed. The fact that it is not 100% reliable shouldn't be basis for restrictions. Even top trained doctors are far from 100% reliable either.