Your question prompts were fluffy and so the answers were fluffy. Would like to see more specific questions. “What are the rates of surgical site infection, recurrence, and bleeding requiring transfusion in open vs lap elective hernia repair?”.
@RichHilsden I worked as an AI developer in Toronto, my wife is a medical graduate preparing for her Canadian Medical License exam, we copied some of the questions asked in MCCQE and I could see, chatGPT solved it perfectly. ChatGPT is based on transformers which google made and open sourced in 2017, Google has its version of ChatGPT called LaMDA and PaLM, it is not open to the public, but Google engineers tested it and thinks it is sentient even now. In 10 years I think this would certainly improve a lot, that’s how computers and software work(Moore’s Law)
You should welcome that. As a primary care doc I want to find ways how to integrate it to better care for patients. It’s just not about making a diagnosis but enhancing the therapeutic relationship and coordination of care.
Precisely, which is why I advise people to compile and save a database of language forms, i.e., questions and commands, which have worked well for them in eliciting good, workable responses from their AI programs. That is critical.
You really seem to care deeply about your profession, as well as helping to educate others. Your videos are definitely appreciated. I do have some questions out of curiosity for a physician like you who really cares about your specialty and being a good doctor. After everything you've done and accomplished up to this point, in a very hypothetical "dream reset" scenario, would you do everything the same all over again while also picking the same specialty? Also going forward, if you were to win some type of mega lottery, would you walk away from the profession? Again, thanks for the videos and discussions. Your content strikes a really good balance of helping both students and the general public.
You are using this so wrong my friend, let me help you out type in "Create a SOAP note for a 66 YO female with PMH of DM2, HTN, and SLE who presents with L MCA stroke s/p TNK on 12/10 currently on ICU day 2. Subjectively she complains of a new mild headache and weakness in her upper and lower extremeties which started this morning, she has no pain or other asociated symptoms. Include normal vital signs and a full physical exam. For the assesment include a detailed differential diagnosis with pros and cons. For the plan include labs and imaging that should be ordered as well as medications with dosing given that she has normal renal and liver function" THEN when it does that type in "make it longer and include pros and cons in the assesment and include a table of a full neuro exam with expected findings based on her chief complaint" If it runs into the character limit type in "keep going" and it will continue. You need to be hyper specific. Think of it this way, all the knowledge is now out there on the web...with google we were using a shovel, with ChatGPT we are using an excavator.
I’ve watched two of this guys videos on ChatGPT but with all due respect, your comment blew me away. This is by far one of single greatest use cases for GPT and no one is talking about it. Thank you for showing me this. What a terrific use
Why this guy don’t make video about this comment. Wow crazy how good it is. I just fry just as you say and it’s amazing. You know you need to know how to ask. If you ask poor it will answer poor. Impressive man
No matter how you ask it, you can't use it's answer in such critical area as a medical care, without verifying it by somebody competent. Please keep in mind these models are made for immitating the human-like conversation and not for providing any exact / trustful analysis - if you get lucky it can give you correct answer (if similar question is included in the training data), but it can also be completely wrong (when it needs to improvize a bit), which is obiviously not acceptable in this case. I strongly suggest anyone to read about how these models work before using it for anything important. You can try this for example: 5+5+5+5+5+3 Now imagine this model counts what the drug dosage should be. When I see how people are using it makes me worried ...
@@jendabekCZ ChatGPT is a DEMO. OpenAI made sure to stress that. It has no access to the internet, it's capabilities are very limited. Wait for GPT-4 to come out with full access to the internet and wait for a fine-tuned model for medicine. That thing will be better than 99% of general practitioners. And it will happen in the next two years.
Umm, why... It'll be way better than doctors. Are you familiar with Iatragenic causes? It's basically death by doctors. It's the 3rd leading cause of death. GPT4 and higher will save lives as it becomes smarter and smarter.
@@ahmerbeg2956 Because these models are guessing the most probable answer, based on the quality of it's training data - there is no actual knowledge of any topic, or a way how to verify whether the response is correct or not. And nobody wants his doctor to be guessing what his treatment / should probably be, right? Without verifying the results (by a competent person or googling the source), nobody should use ChatGPT to anything serious where misleading / incorrect answer is not tolerable. GPT-4 will be very impressive for sure, but the limits will be the same.
@Jan Kadeřábek That's what doctors do, take their best guess. Only they generally don't have good troubleshooting skills (just look at last couple years how wrong they were) or have time to assess all the current data sets on a disease. GPT should be used together to help ensure the doctor is not causing more harm and contributing to Iatragenic causes. Using pattern recognition on test results with AI will help pinpoint any problem. As the test datasets get bigger, the more accurate and smarter it gets.
@@ahmerbeg2956 let's not forget that the AI can only work if you feed it concise information gathered through an anamnesis that could only be reliably done by a medical professional. You try and let this AI give you such a nice answer from the sea of words that can be a 1-on-1 consultation with a patient with no medical knowledge whatsoever.
even if chatgpt answers were limited to google results, there is a major difference... u need to be very specific and need to go through atleast 5 to 10 sites to get a proper answer if u can not write the search queries to the point... with chatgpt its not needed, it learns with ur previous interaction and what u possibly could mean and gives best result possible.. u are a doctor u are asking high standard questions, u are making questions knowing the answer.. for patients its completely opposite we may not even understand ourselves what we are facing or feeling.. plus for the point where u said it seems like its just saying what a doctor may have said in a video or so.. u also learn from ur seniors, books, and other.. at the end u are also repeating others words... finally u said the answer seemed vague or incomplete tupe of thing.. doctors need 10 20 pages of medical history and records and still have difficulty answering things, what are u expecting from a 2 line question
Chat GPTs' answer; You may be referring to the Bible code, also known as the Torah code, which is a theory that suggests that there is a hidden code or message in the text of the Bible. Proponents of the Bible code claim that this code reveals hidden predictions and information about events and individuals, including future events. The theory is based on the idea that if you take the original Hebrew text of the Torah (the first five books of the Bible) and skip certain letters, you can uncover a hidden code or message. Proponents of the Bible code claim that this code reveals hidden predictions and information about events and individuals, including future events. However, the validity of the Bible code has been widely discredited by scholars and mathematicians. They have found that the supposed patterns and predictions are simply the result of selectively searching the text for patterns and using statistical manipulation to make the results appear significant. In conclusion, the Bible code is a controversial and unsupported theory that lacks credible evidence. While it may be intriguing to some, it is not considered a valid scientific or biblical theory by the majority of scholars and experts in the field.
@@threesomeink I That's not the only code. There's others that are lesser known There is a book by Mike Hoggard called KJV Bible numerics. There's a free software that allows anyone to count number of times a word or word phrase is used. It counts for verses and chapters too. Sorry I can't explain well now. There's others codes talked about by Chuck Missler. I've been looking into it for a while now. I did read the Torah Code 25 years ago. This phenomenal KJV Bible code is even more fantastic. It's supernatural.
@@threesomeink also I think it's unlikely anyone would admit that the Torah Code is legitimate. People will deny anything that makes them uncomfortable. That's why this little known phenomena in the KJV Bible code. It takes a long time to get understanding and one won't spend time unless they already believe it. Btw did you read the Torah Code? Or did you just accept what the critics announced to the world? Do you think the official story of 911is true?.. or the lies told by health authorities two years ago were true? Or the official story of the origin of man? Or that life evolved without a God? Or that man built the pyramids without some lost technology like what was used at coral castle in Florida by Ed Leadskalin? My point is that people who believe in something will believe in something else more easily. People who are skeptics may be right but maybe wrong and if you don't believe in one conspiracy or another, if you don't believe in God or whatever, then you likely will find a reason not to believe in anything. There's a lot of people who set out to disprove something and later become a believer. I can think of several off hand. Often the majority of scholars are wrong. That's all. Shalom
@@richardvass1462 Okay but how would counting how many times a word was used in an entire book give you any information other than the amount of times it was used?
I'm a second year medical student. This is just mind boggling. I honestly think it is the greatest single piece of innovation since the internet. It is a fantastic tool, it helps with presentations, it helps with essays, it has been spot on with a lot of clinical cases I have fed it. Complete with diagnosis, treatment, conduct, expected results from interventions. Of course you always have to double check every information it gives you since it's not 100% accurate 100% of the time. But still, just fantastic.
Ask it to 'become a doctor and ask and respond to questions to help you diagnose an issue'. Asking it to become something opens up more possibilities
Your question prompts were fluffy and so the answers were fluffy. Would like to see more specific questions. “What are the rates of surgical site infection, recurrence, and bleeding requiring transfusion in open vs lap elective hernia repair?”.
I am a primary care physician and I see this taking over most of the role of my job within 10 years, I’d be a fool not to see that
Maybe not take over. But will change practice. The internet certainly did.
@RichHilsden I worked as an AI developer in Toronto, my wife is a medical graduate preparing for her Canadian Medical License exam, we copied some of the questions asked in MCCQE and I could see, chatGPT solved it perfectly. ChatGPT is based on transformers which google made and open sourced in 2017, Google has its version of ChatGPT called LaMDA and PaLM, it is not open to the public, but Google engineers tested it and thinks it is sentient even now. In 10 years I think this would certainly improve a lot, that’s how computers and software work(Moore’s Law)
You should welcome that. As a primary care doc I want to find ways how to integrate it to better care for patients. It’s just not about making a diagnosis but enhancing the therapeutic relationship and coordination of care.
I want to talk to a real human not a screen
AI will definitely decrease primary care job openings
You can get it to provide more specific answers. Its a matter for putting in the proper prompts.
Precisely, which is why I advise people to compile and save a database of language forms, i.e., questions and commands, which have worked well for them in eliciting good, workable responses from their AI programs. That is critical.
You really seem to care deeply about your profession, as well as helping to educate others. Your videos are definitely appreciated. I do have some questions out of curiosity for a physician like you who really cares about your specialty and being a good doctor.
After everything you've done and accomplished up to this point, in a very hypothetical "dream reset" scenario, would you do everything the same all over again while also picking the same specialty? Also going forward, if you were to win some type of mega lottery, would you walk away from the profession?
Again, thanks for the videos and discussions. Your content strikes a really good balance of helping both students and the general public.
I like his passion for his career choice as well. He also seems very patient with his patients when explaining any concerns, answering questions, etc
You are using this so wrong my friend, let me help you out type in "Create a SOAP note for a 66 YO female with PMH of DM2, HTN, and SLE who presents with L MCA stroke s/p TNK on 12/10 currently on ICU day 2. Subjectively she complains of a new mild headache and weakness in her upper and lower extremeties which started this morning, she has no pain or other asociated symptoms. Include normal vital signs and a full physical exam. For the assesment include a detailed differential diagnosis with pros and cons. For the plan include labs and imaging that should be ordered as well as medications with dosing given that she has normal renal and liver function"
THEN
when it does that type in "make it longer and include pros and cons in the assesment and include a table of a full neuro exam with expected findings based on her chief complaint"
If it runs into the character limit type in "keep going" and it will continue.
You need to be hyper specific. Think of it this way, all the knowledge is now out there on the web...with google we were using a shovel, with ChatGPT we are using an excavator.
Amazing idea. Love it!
I’ve watched two of this guys videos on ChatGPT but with all due respect, your comment blew me away. This is by far one of single greatest use cases for GPT and no one is talking about it. Thank you for showing me this. What a terrific use
Why this guy don’t make video about this comment. Wow crazy how good it is. I just fry just as you say and it’s amazing. You know you need to know how to ask. If you ask poor it will answer poor. Impressive man
No matter how you ask it, you can't use it's answer in such critical area as a medical care, without verifying it by somebody competent.
Please keep in mind these models are made for immitating the human-like conversation and not for providing any exact / trustful analysis - if you get lucky it can give you correct answer (if similar question is included in the training data), but it can also be completely wrong (when it needs to improvize a bit), which is obiviously not acceptable in this case.
I strongly suggest anyone to read about how these models work before using it for anything important.
You can try this for example: 5+5+5+5+5+3
Now imagine this model counts what the drug dosage should be.
When I see how people are using it makes me worried ...
@@jendabekCZ ChatGPT is a DEMO. OpenAI made sure to stress that. It has no access to the internet, it's capabilities are very limited. Wait for GPT-4 to come out with full access to the internet and wait for a fine-tuned model for medicine. That thing will be better than 99% of general practitioners. And it will happen in the next two years.
2:45 am i watching a Theranos ad, lol?
I hope no pacient will get a treatment completely hallunicated by this LANGUAGE MODEL.
Me too. But the world is changing
Umm, why... It'll be way better than doctors. Are you familiar with Iatragenic causes? It's basically death by doctors. It's the 3rd leading cause of death. GPT4 and higher will save lives as it becomes smarter and smarter.
@@ahmerbeg2956 Because these models are guessing the most probable answer, based on the quality of it's training data - there is no actual knowledge of any topic, or a way how to verify whether the response is correct or not.
And nobody wants his doctor to be guessing what his treatment / should probably be, right?
Without verifying the results (by a competent person or googling the source), nobody should use ChatGPT to anything serious where misleading / incorrect answer is not tolerable.
GPT-4 will be very impressive for sure, but the limits will be the same.
@Jan Kadeřábek That's what doctors do, take their best guess. Only they generally don't have good troubleshooting skills (just look at last couple years how wrong they were) or have time to assess all the current data sets on a disease. GPT should be used together to help ensure the doctor is not causing more harm and contributing to Iatragenic causes.
Using pattern recognition on test results with AI will help pinpoint any problem. As the test datasets get bigger, the more accurate and smarter it gets.
@@ahmerbeg2956 let's not forget that the AI can only work if you feed it concise information gathered through an anamnesis that could only be reliably done by a medical professional. You try and let this AI give you such a nice answer from the sea of words that can be a 1-on-1 consultation with a patient with no medical knowledge whatsoever.
No, I i wouldn't want this to take place of your job.. I need you to stay in business 👍🙂
just u wait till u try gpt4 or gemini or copilot, the choices expand!
Humans and AI need to become one.
Oh yeah and create a supernatural concoction to treat hernia by ultrasound or something like that
I asked chatGPT in front of my patient and it gave me the same hernia information.
even if chatgpt answers were limited to google results, there is a major difference... u need to be very specific and need to go through atleast 5 to 10 sites to get a proper answer if u can not write the search queries to the point... with chatgpt its not needed, it learns with ur previous interaction and what u possibly could mean and gives best result possible.. u are a doctor u are asking high standard questions, u are making questions knowing the answer.. for patients its completely opposite we may not even understand ourselves what we are facing or feeling.. plus for the point where u said it seems like its just saying what a doctor may have said in a video or so.. u also learn from ur seniors, books, and other.. at the end u are also repeating others words... finally u said the answer seemed vague or incomplete tupe of thing.. doctors need 10 20 pages of medical history and records and still have difficulty answering things, what are u expecting from a 2 line question
I liked the video, but I have to say I'm pissed about the clicbait title...
Click baiting huh? Chatgpt didn't school him.
I learned something… maybe it schooled me like an underfunded education system? 😏
I am not having surgeries anymore I am scared
I have scarier videos, you should watch them☠️
@@RichHilsden I won’t, I don’t want more fears
It is unreliable because it has no access to the internet, books online, publications...etc etc
soon. very very soon
@@jsejjd hope so.
It is essentially in the test phase right now
I've learned there is a Bible numeric code in the KJV Bible which has no possibility of being random. Could you ask this chat bot about it please?
Chat GPTs' answer;
You may be referring to the Bible code, also known as the Torah code, which is a theory that suggests that there is a hidden code or message in the text of the Bible. Proponents of the Bible code claim that this code reveals hidden predictions and information about events and individuals, including future events.
The theory is based on the idea that if you take the original Hebrew text of the Torah (the first five books of the Bible) and skip certain letters, you can uncover a hidden code or message. Proponents of the Bible code claim that this code reveals hidden predictions and information about events and individuals, including future events.
However, the validity of the Bible code has been widely discredited by scholars and mathematicians. They have found that the supposed patterns and predictions are simply the result of selectively searching the text for patterns and using statistical manipulation to make the results appear significant.
In conclusion, the Bible code is a controversial and unsupported theory that lacks credible evidence. While it may be intriguing to some, it is not considered a valid scientific or biblical theory by the majority of scholars and experts in the field.
@@threesomeink I
That's not the only code.
There's others that are lesser known
There is a book by Mike Hoggard called KJV Bible numerics.
There's a free software that allows anyone to count number of times a word or word phrase is used. It counts for verses and chapters too. Sorry I can't explain well now.
There's others codes talked about by Chuck Missler. I've been looking into it for a while now. I did read the Torah Code 25 years ago. This phenomenal KJV Bible code is even more fantastic. It's supernatural.
@@threesomeink also I think it's unlikely anyone would admit that the Torah Code is legitimate. People will deny anything that makes them uncomfortable.
That's why this little known phenomena in the KJV Bible code. It takes a long time to get understanding and one won't spend time unless they already believe it. Btw did you read the Torah Code? Or did you just accept what the critics announced to the world? Do you think the official story of 911is true?.. or the lies told by health authorities two years ago were true? Or the official story of the origin of man? Or that life evolved without a God? Or that man built the pyramids without some lost technology like what was used at coral castle in Florida by Ed Leadskalin? My point is that people who believe in something will believe in something else more easily. People who are skeptics may be right but maybe wrong and if you don't believe in one conspiracy or another, if you don't believe in God or whatever, then you likely will find a reason not to believe in anything. There's a lot of people who set out to disprove something and later become a believer. I can think of several off hand. Often the majority of scholars are wrong. That's all. Shalom
@@richardvass1462 Okay but how would counting how many times a word was used in an entire book give you any information other than the amount of times it was used?
✋ Promo`SM!!
Privacy, security.
Chat gpt is start of normality
Cheaper, faster, and better.....
I wonder how good a tool this would be for medical students. Maybe an idea for another video.
I'm a second year medical student. This is just mind boggling. I honestly think it is the greatest single piece of innovation since the internet. It is a fantastic tool, it helps with presentations, it helps with essays, it has been spot on with a lot of clinical cases I have fed it. Complete with diagnosis, treatment, conduct, expected results from interventions. Of course you always have to double check every information it gives you since it's not 100% accurate 100% of the time. But still, just fantastic.