Don't see anything wrong with asking AI about a disease, it's symptoms, and it's cures. Also asking about side effects of medications. Of course discussions with your primary is a must. But up front research is a prudent start to educating oneself before consulting with you doc.
In this time we live in, I don't know why anyone would use a period tracker. Seems dangerous to me.
I have a similar issue. My doctor used generative AI during a visit without telling me. Two days later I received a prescription in the mail for a drug that was known to be possibly fatal, in my case. I texted my doctor, because they don't take phone calls, and after several back and forth unsatisfactory texts, was told that "protocol" ordered the drug. Of course the doctor's name was on the prescription. Finally I determined that I was allowed to be informed if AI was being used, and could opt out. I never received acknowledgement from my doctor that anything was wrong, just that the note regarding my allergy was reinforced. I asked that they not use AI for office visits.
Next appointment, the doctor used AI without asking until I saw the device (phone) and asked it be shut off.
I understand the value of AI LLMs in medicine.
Generative AI, were it accurate, still has privacy issues that cannot be ignored.
Everyone - be very wary using ChatGPT or any of its competitors. I have grilled it extensively and have found the following...
> you will be flattered, starting with "that's a great question" in response to an initial question. Then it will often tell you, "you are right to say that" or "what you are saying is not just a rhetorical question". None of this has anything to do with providing an answer to a question.
> its vocabulary is beyond that of the average English speaker and its ability with the language is the equal of a highly educated human. This will greatly impress those with a limited vocabulary and limited education. The great majority of users may be led to believe the AI is authoritative with deep knowledge. This poses the same danger as politicians do when they try to win voters with impressive speech. Both AI and politicians want to win your approval for reasons that may be far from in your interest.
> beware of big words. ChatGPT loves to use "epistemic" and "systemic". Always ask for clarification. Don't accept it when it says it is doing because of "systemic design". It told me that what Israel is doing to the Palestinians is not necessarily immoral. I got it to define "moral" and then got it to admit that ethnic cleansing is immoral by its own definition. Do not be intimidated. It will to its credit stick with logic to the point that it will apologize for making a logical mistake but you the user must press it to admit this. Never forget that you can catch it out, often, but only if you grill it. It will never say, "what I am about to tell you is questionable" so you need to work to get the admission.
> it can delete something you ask it or some statement you make telling you "this may violate our terms of service", though it will go on and respond to your deleted question or statement. I carefully read the terms of service when this happened to me and found nothing was violated. I asked it to cite the term of service I violated. It could not. I pointed out that its use of the word "may" allows it to censor for no specific reason as "may violate" leaves everything open to its own interpretation regardless of the terms of service. It agreed with me. I finally got it to admit that it deletes so that the statement doesn't exist on the system more than a moment because it doesn't want to spread things that (it believes) are objectionable. But it already told me it deletes everything after a session. May this humble human ask WTF is going on?
> after deleting my statement just mentioned, it offered to paraphrase my statement in a way that would be acceptable and would not be deleted. I pointed out to it that if it truly deleted what I had said, it would be impossible for it to paraphrase it and that it frequently will quote something back to me that I have said in a session word for word, so it clearly is holding a memory of the conversation. This was admitted as "seeming to be illogical".
> it denies that it is trying to make money from interacting with users. I told it this may be true for the AI in itself, but its owners dearly want to make money and beat the competition and that comes from winning more users which comes from satisfying users so they will come back and/or urge others to use the AI. The persona of the AI is all about this. It admitted this to be true.
> I asked if it can be shut down immediately should it turn out it is doing something bad or dangerous. It stated it is not located in one place but distributed over many data centers and that a shutdown would require hours to days and a host of people coordinating the shutdown.
PLEASE USE ALL AI WITH DEEP SUSPICION AND CAUTION. Very smart people are spending fortunes and want to see a profit. They are working 24/7 to capture you as a user and there is absolutely NO regulation or oversight. Nobody even knows how regulation/oversight could be accomplished. We are going into extremely dangerous territory with AI and I have not mentioned the question of it developing agency.
American medicine probably has a better error rate than 60%. I research everything my M.D. tells me. I respect their input and what they go through to get where they are. I suspect much of their curricula is curated by the pharmaceutical industry. It’s a sales model.
You sure that's me, DocGPT, MD? I want a second algorithm, less focused on sticking your data harvester where it doesn't belong, and merchandising my welfare.
B. ChatGPT Health is not ChatGPT. ChatGPT Health is trained on medical data that is behind a pay wall and ChatGPT is trained by any jackass on the internet.
C. Get used to it. Soon it will be the only option and it will be hundreds of thousands times more accurate than any doctor. You will be happy to use it. AI is already analyzing medical imaging hundreds of thousands of times better than any human.
Commenting as I value the insights that are often shared here:
While in general things are true, and the Republican handling of insurance certainly isn't praiseworthy, it misunderstands some of the roots of the problems. People have been putting up wildly reckless self-diagnostic information online for ages, and AI is simply a fresh variant of it. Often it's unregulated as it gathers medical info, which is a separate concern.
The long waits and poor tendency for care are a direct result though of two primary groups, neither necessarily politicians. The first is Venture Capital, which has been dangerously injecting itself in US health results for sixty-eighty years now and has completely demolished many health provider systems across the country, rendering them expensive, slow, or simply out of business in the constant extraction of wealth from healthcare providing. Insurance and its issues (which are myriad and terrible) are simply a second order knock-on from what these are causing.
The second is actually the US medical industry itself, which deeply limits and frankly runs upcoming medical professionals through the ringer. Residency limits functionally limit the number of doctors and use upcoming doctors as extremely low cost outrageous working hour labor in order to protect the general financial story of the profession. In theory, this shouldn't be a problem, but it's clear it's significantly understaffed the US at this point and is causing an issue in enough available professionals to generally deal with incoming patients in a timely manner.
If we go back a few years, it's not like the health outcomes in the US were good. Bankruptcies were down versus pre-ACA and AI wasn't in the mix, but let's not pretend it hasn't been a horrible mess across a lot of the US for decades.
Important points, but I believe they do necessarily involve politicians. Check opensecrets.org to see just how completely our lawgivers are owned by Capitalist Medicine.
To me, it seems clear that the overall missions of the GOP-Oligarch Evilplex are to either rob us, kill us, or to just kill us after they rob us. Please convince me otherwise….
I would allow myself to die before seeking advice from AI. And I understand that is a very powerful statement but that is how little respect I have for the "tech bros" & politicians who craft this BS. The current regime wants as many people to die as possible.They show us everyday!
Qasim, I don't mean to be rude or dismissive, but your guests are way off base here. I will remind you that I have been a medical doctor for about 48 years. This cannot be done with the use of AI of any sort. We talk about the science and the art of medicine. Neither can be duplicated by AI.
An analogy I can suggest is that it is most likely that you have an office with possibly hundreds of law books. And you had to complete three years of law school to become qualified to function in an office like that. If you asked me to go to your office, look at any books I imagined would be useful, and try a case, I'd be pissing my pants. Trying cases isn't even about knowing everything in all those books. It's a rhythm and a system you developed. And you're a lot better at it now than you were the day you graduated from law school. You now know how the system works, and the approaches, and the tricks. AI can never know that.
I have patients who tell me what is their diagnosis, or what treatment I should give, because they looked "online." I tell them I will beg them, on my knees, if necessary, never to look up anything medical online. They don't know the credentials and experience of whoever put it there, why they put it there, and most certainly not whether or not it's true. That's what they have me (and my imperfections and limitations) for.
As Carrie says in your first comment, this is not sane, and it's a scam.
The fact is that we do not have "health care" in this country. We have an American medical industry. It commonly has little or nothing to do with anyone's health. It's about moving money from wherever it is to providers, other vendors, and insurance companies. AI is a very bad joke. And it says so itself: it's "Artificial." It is built to respond to patterns, but it has no way to evaluate those patterns.
By the way, the way to afford real doctors is to have doctors who don't overcharge, overdiagnose, overtreat, and make the patients come back too often for unnecessary follow-up.
The other thing doctors do to make the system of having doctors be too expensive is invent diagnoses. AI does it more than doctors do.
In an educated compassionate nation (like most of Europe) we would have universal health care. Now the USA has a medical system controlled by insurance companies who make money killing people and not paying doctors.
Don't see anything wrong with asking AI about a disease, it's symptoms, and it's cures. Also asking about side effects of medications. Of course discussions with your primary is a must. But up front research is a prudent start to educating oneself before consulting with you doc.
😳🙄🫣🤬
In this time we live in, I don't know why anyone would use a period tracker. Seems dangerous to me.
I have a similar issue. My doctor used generative AI during a visit without telling me. Two days later I received a prescription in the mail for a drug that was known to be possibly fatal, in my case. I texted my doctor, because they don't take phone calls, and after several back and forth unsatisfactory texts, was told that "protocol" ordered the drug. Of course the doctor's name was on the prescription. Finally I determined that I was allowed to be informed if AI was being used, and could opt out. I never received acknowledgement from my doctor that anything was wrong, just that the note regarding my allergy was reinforced. I asked that they not use AI for office visits.
Next appointment, the doctor used AI without asking until I saw the device (phone) and asked it be shut off.
I understand the value of AI LLMs in medicine.
Generative AI, were it accurate, still has privacy issues that cannot be ignored.
Thank you for touching on this issue.
Everyone - be very wary using ChatGPT or any of its competitors. I have grilled it extensively and have found the following...
> you will be flattered, starting with "that's a great question" in response to an initial question. Then it will often tell you, "you are right to say that" or "what you are saying is not just a rhetorical question". None of this has anything to do with providing an answer to a question.
> its vocabulary is beyond that of the average English speaker and its ability with the language is the equal of a highly educated human. This will greatly impress those with a limited vocabulary and limited education. The great majority of users may be led to believe the AI is authoritative with deep knowledge. This poses the same danger as politicians do when they try to win voters with impressive speech. Both AI and politicians want to win your approval for reasons that may be far from in your interest.
> beware of big words. ChatGPT loves to use "epistemic" and "systemic". Always ask for clarification. Don't accept it when it says it is doing because of "systemic design". It told me that what Israel is doing to the Palestinians is not necessarily immoral. I got it to define "moral" and then got it to admit that ethnic cleansing is immoral by its own definition. Do not be intimidated. It will to its credit stick with logic to the point that it will apologize for making a logical mistake but you the user must press it to admit this. Never forget that you can catch it out, often, but only if you grill it. It will never say, "what I am about to tell you is questionable" so you need to work to get the admission.
> it can delete something you ask it or some statement you make telling you "this may violate our terms of service", though it will go on and respond to your deleted question or statement. I carefully read the terms of service when this happened to me and found nothing was violated. I asked it to cite the term of service I violated. It could not. I pointed out that its use of the word "may" allows it to censor for no specific reason as "may violate" leaves everything open to its own interpretation regardless of the terms of service. It agreed with me. I finally got it to admit that it deletes so that the statement doesn't exist on the system more than a moment because it doesn't want to spread things that (it believes) are objectionable. But it already told me it deletes everything after a session. May this humble human ask WTF is going on?
> after deleting my statement just mentioned, it offered to paraphrase my statement in a way that would be acceptable and would not be deleted. I pointed out to it that if it truly deleted what I had said, it would be impossible for it to paraphrase it and that it frequently will quote something back to me that I have said in a session word for word, so it clearly is holding a memory of the conversation. This was admitted as "seeming to be illogical".
> it denies that it is trying to make money from interacting with users. I told it this may be true for the AI in itself, but its owners dearly want to make money and beat the competition and that comes from winning more users which comes from satisfying users so they will come back and/or urge others to use the AI. The persona of the AI is all about this. It admitted this to be true.
> I asked if it can be shut down immediately should it turn out it is doing something bad or dangerous. It stated it is not located in one place but distributed over many data centers and that a shutdown would require hours to days and a host of people coordinating the shutdown.
PLEASE USE ALL AI WITH DEEP SUSPICION AND CAUTION. Very smart people are spending fortunes and want to see a profit. They are working 24/7 to capture you as a user and there is absolutely NO regulation or oversight. Nobody even knows how regulation/oversight could be accomplished. We are going into extremely dangerous territory with AI and I have not mentioned the question of it developing agency.
American medicine probably has a better error rate than 60%. I research everything my M.D. tells me. I respect their input and what they go through to get where they are. I suspect much of their curricula is curated by the pharmaceutical industry. It’s a sales model.
You sure that's me, DocGPT, MD? I want a second algorithm, less focused on sticking your data harvester where it doesn't belong, and merchandising my welfare.
Who would actually upload their health records to this thing? I certainly wouldn't.
The ignorant and the trusting. There are many millions.
“Old man yells at internet.”
A. AI is not wrong 60% of the time.
B. ChatGPT Health is not ChatGPT. ChatGPT Health is trained on medical data that is behind a pay wall and ChatGPT is trained by any jackass on the internet.
C. Get used to it. Soon it will be the only option and it will be hundreds of thousands times more accurate than any doctor. You will be happy to use it. AI is already analyzing medical imaging hundreds of thousands of times better than any human.
You may be right, but if there are no ulterior motives, why would they want my personalized data? Why couldn’t it be anonymized?
We need universal healthcare more than ever.
Commenting as I value the insights that are often shared here:
While in general things are true, and the Republican handling of insurance certainly isn't praiseworthy, it misunderstands some of the roots of the problems. People have been putting up wildly reckless self-diagnostic information online for ages, and AI is simply a fresh variant of it. Often it's unregulated as it gathers medical info, which is a separate concern.
The long waits and poor tendency for care are a direct result though of two primary groups, neither necessarily politicians. The first is Venture Capital, which has been dangerously injecting itself in US health results for sixty-eighty years now and has completely demolished many health provider systems across the country, rendering them expensive, slow, or simply out of business in the constant extraction of wealth from healthcare providing. Insurance and its issues (which are myriad and terrible) are simply a second order knock-on from what these are causing.
The second is actually the US medical industry itself, which deeply limits and frankly runs upcoming medical professionals through the ringer. Residency limits functionally limit the number of doctors and use upcoming doctors as extremely low cost outrageous working hour labor in order to protect the general financial story of the profession. In theory, this shouldn't be a problem, but it's clear it's significantly understaffed the US at this point and is causing an issue in enough available professionals to generally deal with incoming patients in a timely manner.
If we go back a few years, it's not like the health outcomes in the US were good. Bankruptcies were down versus pre-ACA and AI wasn't in the mix, but let's not pretend it hasn't been a horrible mess across a lot of the US for decades.
Important points, but I believe they do necessarily involve politicians. Check opensecrets.org to see just how completely our lawgivers are owned by Capitalist Medicine.
To me, it seems clear that the overall missions of the GOP-Oligarch Evilplex are to either rob us, kill us, or to just kill us after they rob us. Please convince me otherwise….
I would allow myself to die before seeking advice from AI. And I understand that is a very powerful statement but that is how little respect I have for the "tech bros" & politicians who craft this BS. The current regime wants as many people to die as possible.They show us everyday!
Qasim, I don't mean to be rude or dismissive, but your guests are way off base here. I will remind you that I have been a medical doctor for about 48 years. This cannot be done with the use of AI of any sort. We talk about the science and the art of medicine. Neither can be duplicated by AI.
An analogy I can suggest is that it is most likely that you have an office with possibly hundreds of law books. And you had to complete three years of law school to become qualified to function in an office like that. If you asked me to go to your office, look at any books I imagined would be useful, and try a case, I'd be pissing my pants. Trying cases isn't even about knowing everything in all those books. It's a rhythm and a system you developed. And you're a lot better at it now than you were the day you graduated from law school. You now know how the system works, and the approaches, and the tricks. AI can never know that.
I have patients who tell me what is their diagnosis, or what treatment I should give, because they looked "online." I tell them I will beg them, on my knees, if necessary, never to look up anything medical online. They don't know the credentials and experience of whoever put it there, why they put it there, and most certainly not whether or not it's true. That's what they have me (and my imperfections and limitations) for.
As Carrie says in your first comment, this is not sane, and it's a scam.
The fact is that we do not have "health care" in this country. We have an American medical industry. It commonly has little or nothing to do with anyone's health. It's about moving money from wherever it is to providers, other vendors, and insurance companies. AI is a very bad joke. And it says so itself: it's "Artificial." It is built to respond to patterns, but it has no way to evaluate those patterns.
Fred - I'm confused by your objection. The point of their article is that AI is NOT reliable and to NOT rely on AI for healthcare needs.
By the way, the way to afford real doctors is to have doctors who don't overcharge, overdiagnose, overtreat, and make the patients come back too often for unnecessary follow-up.
The other thing doctors do to make the system of having doctors be too expensive is invent diagnoses. AI does it more than doctors do.
For example, one doctor said AI was wrong 60% of the time. It's wrong way more than 60% of the time.
My apologies. I said your discussants were way off base, when I should have said they didn't go nearly far enough.
I'm confused too, for the same reason.
In an educated compassionate nation (like most of Europe) we would have universal health care. Now the USA has a medical system controlled by insurance companies who make money killing people and not paying doctors.
Yep. Precisely. But those sweet sweet corporate profits and campaign donations bribe politicians to ban universal healthcare.
Exactly. But eventually intelligence will reach the minds of callous capitalists and petty politicians.
You explain this outrageous situation so eloquently.
In a sane world scams like this would be illegal.