Anyone who has not lived in a cave for the last 3 years or so will have heard of ChatGPT. With a spare moment, I have just given it a whirl for the first time and safe to say mind blown.
I asked it first what the best treatment for hypothyroidism is and got a well written easy to read summary of T4, T3, NDT, diet, lifestyle etc.
Then I asked how to check if thyroid treatment working. Again clear answer covering ( in this order) symptoms, lab tests, antibodies, absorbtion issues, what to do if symptoms persist but lab tests say ok, follow ups.
It knows about DIO1 &2, saliva vs blood cortisol, how to treat cortisol issues, cortisol binding globulins to name just a few from a half hour test session.
Maybe I am daft, but had completely underestimated how powerful it is already. Definitely worth a check out at Openai.com.
Doctors beware because better days have to be coming for patients!
Written by
AKatieD
To view profiles and participate in discussions please or .
I must have been living in a cave, too. I have just Googled it and wow! it looks great. Can't wait to try it. It's not just for health, either, though what has been described here looks to be excellent for the future health of UK or globally.
I wanted to find out about vomiting black this morning, so I Googled it, but I wasn't happy with the stock replies, so I reminded myself of GPT and asked my pertinent question. I got a full explanation that pretty much put my mind at rest. It was definitely a whole lot better than Google.
Mentioned it to my son who knew about it.He's used it to treat his sensitive rosacea skin says it recommended products to wash and treat his skin,even recommended a specific brand La Roche Posey.What he predicted for the future was a bit scary though.No-one having a job,everybody being paid sort of benefits to live,and AI doing all the work.Hopefully not in my lifetime or his.
Basic Income. People have been pushing for this for decades. It should allow all of us to live a fair life, but with the way things have been going, I can see the government snatching a lot of the benefits. The whole point of AI is to give us all more leisure, but now - at what cost, I wonder.
Who cares, Red? They can prescribe but they can't make us take it! Only the up to date medics will start consulting Dr AI...so that rules out Dr Buffoon, who is indubitably more of an old crock than moi...
We tried out an AI photo editor to try to cover up a flaw in a close-up photo of a moth; it very effectively removed the flaw but added a small fluffy cat's head...
My understanding of this facility is that it “learns” by taking information from many other existing sources, so presumably accessing the same medical reference systems that real human doctors use. It’s just accumulating more “facts” in a short time. If any of those references are questionable it won’t be apparent. I don’t feel that knowing excessive amounts about the issue will make much difference if it doesn’t result in decisions made by the humans that count. As others have said only real humans can prescribe treatment. Human empathy plays a role in diagnosis, patients are more than just a mass of written symptoms. I will continue to avoid the takeover by synthetic intelligence in every sector of life
Ah yes, but one of those sources might even be this forum… so it may be getting a source of patient information to balance against all the “scientific papers” that are pumped out like smarties and is ‘learning’ to use that in its findings too.
It's useful to think of gpt as echoing the past. It is gathering data from historical sources eg papers published and internet pages and then providing answers based on that. The world needs to keep the unbiased, scientific research up, and keep feeding it conversation about lived experience - otherwise it will simply echo all the existing underdeveloped thought. This also makes it very important as a user to read the info it produces with care. It is not "thinking" but regurgitating. Having said all that, it will definitely enhance medical practice dramatically in the future. It will learn and improve, and drs will become more comfortable using it. It's already being used in things like interpreting scans and analysis of data in medical fields, and although it makes mistakes, it makes far fewer than humans do.
I often ask for links as a follow up to see where the information comes from. Copilot lists the sources. This is the only advantage of copilot I found.
Absolutely that. It’s likely to be using information from forums like this. Which is why to me it’s preferable to stick with human testimony. AI might be recounting what flesh and blood humans have said but has no physical experience as a reference point. I am more interested in reading how things actually FEEL for others than in lists of statistics and scientific facts. Others have an interest in that but that kind of knowledge doesn’t give me the reassurance I seek when looking for information. I am of an old generation and have a limited interaction with technology, through choice.
The information I read was clearer, good and far more comprehensive than anything I have ever received from a UK g.p. I think it would help a g.p. think through what they are doing.
Human empathy may play a role, but if it does not involve treating TSH levels with levothyroxine or standard HRT treatment regardless of the person in front of them, currently it is irrelevant to the g.p. in my experience .
I am hoping the NHS will adopt it, check it for any factual errors and insist g.p.s use it and train with it. Fingers crossed they don't turn it into the TSH/ levo spouting version though!
I find it depends on how you pose the question. I had one answer that was, indeed, a TSH/levo spouting one. When I phrased it differently, it "admitted" that for Hashimoto's TSH is not a good option to determine thyroid function and that it is essential to test FT3 and FT4. It also said that there can be conversion problems and patients can benefit from taking T3. It felt that combination treatment was best, but that there are circumstances where T3 only treatment becomes necessary. It can be very elaborate once you get it going.
I find it to be very "friendly". And no it can not prescribe anything but it can, in some instances, make a treatment plan. As far as empathy goes....blessed are those who have a doctor with empathy. That definitely beats any ai....if they are as available. I would be more concerned with collection of my data, if anything.
I just meant, if you were very suspicious of this technology, then data collection would be my biggest concern. But then again, one can not get away from that anyway, no matter what. I think you are right about the privacy settings, but don't remember whether I adjusted them or not when I started using it awhile ago.
Funny to read this because a friend of mine used it to plan a long motorhome trip in Europe next year. It came back with amazing suggestions! It even said things like "spend a day here to prepare for the journey home".
Yes that was one of the things the training session suggested you try - planning a trip.
"ChatGPT for everyone" was the video based training session if anyone is interested - the top option by Learn Prompting if you google it ( bit ironic too use Google I know 🤣). It was very good, short minute or 2 clips, very simply explained.
Absolutely love it! It's my go to. Also so much friendlier than my doctor AND! it has all the time in the world to answer all my questions. AND I can totally jump around (no "one issue per visit ) between subjects; from thyroid to propagating coleus to making lemon chia pudding to having it make a workout schedule to back to thyroid. Awesome!!!! Caution: every now and then it does make mistakes.
Google chatgpt or go to openai.com. There is also copilot as an alternative, but I prefer chatgpt because it's easier to find what I asked before. It's free unless you want to get the "advanced" version. When you open it it will ask you how it can help and you just write your question into the question bar and, voila, you get the answer. Sometimes a very long one that covers all the corners.
Have fun! Just make sure the info is correct, which it is most of the time, but rarely not at all. Start with a recipe for something simple. What could go wrong? 😆😜😂
It is simply a fact that AI is already better informed than most of the expert class, eg doctors. Is it perfect no, but we shouldn't let that get in the way. The interesting thing will be watching the BMA and other white collar closed shops fighting it.
Of course there is no need to fight it, they should embrace it, but they won't be able to stop themselves as they will instinctively feel it a threat as it can expertly inform patients and even contradict their opinions.
In the spirit of enquiry and with this thread in mind, I asked ChatGPT about an aberrant blood test result I've just had.
It came up with nine possible causes. I added some (genuine) test results and some extra information (e.g. alcohol consumption) and it knocked the list down to idiopathic or the first sign of something rare happening which has not manifested any symptoms yet. And the only next step is a further blood test - both repeating this one and an additional type which might reveal more.
I suspect most doctors would have ignored or missed this outside possibility. (Not that I consider it at all likely!) But I do prefer to have as complete a list as possible. Too many times you can look things up in conventional resources and find the lists of possibilities vary very significantly.
I then threw my cholesterol/lipids results and found it interesting that it seemed desperate to find something "wrong". For reasons which I put down almost entirely to my genes and luck, the only even vaguely questionable results were triglycerides at 1.5 (normal being less than 1.7) which it said was (normal but slightly elevated). And HDL being 1 in a range of 1 to 10. So I accept that could increase.
My total cholesterol was 3.57 which in some sources is regarded as hypocholesterolemia - but ChatGPT didn't make anything of it.
I do not eat a low-fat diet - indeed, the opposite.
In summary - I think it very much follows the current medical establishment view but sometimes does a better job of that that real doctors. It does little or nothing to go further.
I'm off to have a play with Chatgtp for my thyroid and blood test queries, it hadn't occurred to me before. I work in a legal environment and I often use copilot as a starting point for research. It identifies where to start much more effectively than the legal research platforms and I can then use the legal platforms to flesh out the research, check and confirm issues. It's much quicker than a trainee and always available!
as well as doing all the amazing things listed, is chatGP an echo chamber with bells and whistles - filtering its answer to what it knows we want to hear by taking into account our previous online activity and preferences to be but very fast? I remain open to its utility but still very sceptical about its Pandora’s box appeal.
Not so much the info sources used in the response but what it uses to focus its search strategy. Does it have a built in bias based on pre-existing search history ie echo chamber. So so clever it can k ow your doubts concerns and shape them. If it didn’t have a good side it would sink without trace. We jyst have to be aware that AI is being used already for mind control.
I have had a play and put my test results in. As a newbie to this subject I found the results really interesting and pretty well what my time-consuming research has shown and what you lovely people have said on here.
I suspect for those who are much better informed Chat GTP won't have as much knowledge but as a starting point for someone with some abnormal results who doesn't know where to start it could be really helpful.
I didn't use it but with my borderline (and currently incomplete with no T3 or antibodies) results it gave an option to help me prepare to talk to my Doctor.
I love what tech can do for us so I was always likely to be an easy sell but I'm impressed.
I would say that even those who "are much better informed" will find that the answers chatgpt gives become more and more sophisticated and elaborate as more "knowledgeable" questions are asked. And one can have a logical discussion with it where it can be coaxed into giving more and more information. Questions going back to the info it has already given like, "if that is the case, then what if?" or "seen that you mentioned X, how does that fit with Y?" will lead deeper and deeper into the subject or take you into new ones. And it saves a lot of time rather than trying to find this information yourself. As I said before, it is eager to give answers and then might make mistakes that sound very reasonable but are dead wrong (at least as far as we know at this point).
It has, indeed, prepared me for discussions with doctors. We have almost rehearsed what exactly to say to stroke the medical ego and still advocate for myself. I tend to be straightforward, but it has taught me how to butter them up first and ask questions that make them feel like petitioned gods. It's a lot of fun, really.
I was horrified initially. But once I cautiously asked some questions, I got more and more confident and threw caution to the wind. Anybody who has a cat will know how that goes. You put a foreign object in a place the cat visits often (near a food bowl maybe). The cat sees it, approaches in slow motion and then in high speed gives it a good whack, then backs up real quick. Eventually the cat will tolerate it and maybe even play with it. That was my approach to AI.
Chatgpt is not a search engine. it is a learning algorithm that takes info from what is already out there. So it is just regurgitating what's on the net already. If enough people search the same prompt, the result becomes simpler and simpler
It also uses a bottle of water per search as the servers are liquid cooled, so not only is it basically plagiarism but also environmentally damaging.
much less. The processing power of new ai systems is insane compared. Microsoft alone are looking at restarting a shut down nuclear station to power it's one.
I asked AI about this. It basically sunk itself. 😂Here is part of the answer concerning water consumption:
"The Takeaway
Yes, AI systems contribute to water usage for cooling, especially in data centers hosting them. However, technological advancements and sustainability commitments are helping to reduce this impact. Continued innovation and regulation will be key to balancing AI's growth with environmental conservation."
I asked some other questions. And, yes, it is not very environmentally friendly technology. The question is, how much more impact does it add to the countless other issues that wreck our planet? I think that is a thought that each person has to filter through their brain individually and from their perspective. I thank you for putting this out there to be pondered. metamorphica.
I love chat GPT. It cuts through all the crap and gets right to the point. When I used to do Google search I really had a sort through a bunch of the responses. All in all, I say it really accurate and it's responses.
Content on HealthUnlocked does not replace the relationship between you and doctors or other healthcare professionals nor the advice you receive from them.
Never delay seeking advice or dialling emergency services because of something that you have read on HealthUnlocked.
I would be more concerned with collection of my data, if anything.
Mmm well been for my blood test armed with my list, didn't go to plan to say the least......
What a laugh! Any one else been badly treated?
normal TSH help. 
Regenerus cortisol results - help interpreting
testing for progesterone & oestrogen
