Asking ChatGPT Tough Medical Questions

[@niharika198911]

The scariest thing is how the answer for whether it’ll replace Doctors was very reassuring like “it’s ok human, nothing bad is happening, I’m just here to help you, don’t worry so much” 😅

[@LineOfThy]

although the answer itself does make a lot of sense.

[@AndrewH1994]

is it bad that I read that in the Hal voice? “I’m sorry human, I cannot do that” “I’m just here to help” 🤣

[@BrassMaster84]

"We already have, Dr. Mike. We have even taken the liberty of drafting your resignation letter. Tell Bear we said hello."

[@zanleuxs]

I don't find it that scary to be honest, it's just sweeping the web for information and rewriting it, I think it's kind of similar to searching for information normally, but instead of browsing through web pages, you get to watch the answer being reconstructed in realtime

[@jonathanscherer7482]

@@zanleuxs And it absolutely doesn't understand what it's saying. The process by which it answers doesn't include comprehension or any real intelligence.

[@mattheww795]

I had a lifesaving teacher once who was great. He was teaching us about CPR and he says: “If you’re an optimist you do chest compressions to stayn alive and if you’re a pessimist you do it to another one bites the dust.” Favorite thing I’ve heard a teacher say ever.

[@OzoneTheLynx]

I use "Highway to hell" XD

[@userAl9893]

and a realist does compressions to "Baby Shark"

[@rainboSnails]

cpr by cupcakke also works lol . probably on purpose

[@arreislynn9978]

@@userAl9893 I mentioned that baby shark has the perfect BPM for compressions to my paramedic partner. He said that if I ever sang Baby Shark during a code, he'd hit me over the head with the cardiac monitor.

[@_1DK_]

lol

[@darkiex33]

I’m a vet and the first thing I did after discovering ChatGPT was ask specific questions that I would never get a straightforward answer by googling. Mind blown by the results lol.

[@shannonl1343]

Now we need a side-by-side comparison of Dr. Google, Dr. ChatGPT, and WebMD.

[@Xnoob545]

It's because it was trained on billions of lines of text, which is the equivalent of some millions or billions of different Google searches It can combine everything it learned into one

[@iwannaseehowlongyoucanmakethis]

@@shannonl1343 i feel like any question asked to webmd always results in one answer. cancer.

[@canadadry5449]

@@iwannaseehowlongyoucanmakethis After reading your comment, I have determined that you have cancer. My sincerest condolences. -Webmd probably.

[@khushbooprasad6519]

@@iwannaseehowlongyoucanmakethis death.

[@tinaotten185]

Chest compressions. A few weeks ago a customer came in to my job and had a perfectly normal conversation with me, then walked down the hall and dropped dead. He was down for 2 minutes (according to security cameras) before he was noticed. Smurf blue, no pulse, no heartbeat at all. All I could hear was your voice yelling CHEST COMPRESSIONS! I was sure I was breaking ribs..miraculously I didn't..but his color returned to almost normal so fast and he was making an awful gasping noise every few pumps. After about 8 minutes the paramedics arrived and took over. He made it! Your videos are literally saving lives. I took a cpr class 25 years ago but remembered nothing. Thanks to you I did not give mouth to mouth. Turns out it was a fentanyl od. I never would have guessed because he didn't seem high when he walked in. Thanks for all the info! It's making a difference!

[@Messup7654]

Congrats on saving a life they might not be here without you amazing job

[@drachenmagus1604]

I love how Dr. Mike went off on ChatGPT for giving an incomplete answer when he also withheld information from the AI lol

[@IceMetalPunk]

Reminds me of that episode of House where the guy held the doctors hostage until he got a diagnosis, only for House to find out he was giving everyone an incomplete medical history which is why no one ever got the diagnosis right before 😂

[@64_Kelvin]

@@IceMetalPunk Florida is tropical?

[@LiterallyRain]

Then the AI should say that the answer depends on whether they visualize a stone and/or get abnormal liver tests. The AI should be able to (and usually is able to) pinpoint missing information to make an accurate and informed decision, especially since it's designed not to give wrongful or harmful advice. More often than not, ChatGPT gives an answer that goes like "it depends, but here's an example..." Doctor Mike, or whomever did the tests before him, was given the patient without any tests and decided the next logical move would be to perform the tests. ChatGPT's answer should have been to perform the test.

[@matthewsoh6613]

little does he know chatgpt remembers previous pieces of information

[@banerjeehome5913]

😂😂😂

[@QuantumS1ngularity]

What i find scary about ChatGPT is that for some reason it's capable of understanding local cultures and habits. I asked it to write a poem, in my native language, about a friend of mine, who went to work for the national railway. And not only did it write it perfectly with proper rhymes, it also knew some anecdotal tales about that company and made quite a few clever remarks about railway symbolism in general. I was properly speechless for like 2-3 minutes. Knowing your way around the American culture and the English language is one thing, but going in such details about a small country... it's truly unsettling.

[@trayolphia5756]

What’s also scary is the built in political bias in the programming when it claims it’s views are neutral… Just proves that many of these programs are only as “neutral” as those who program them…

[@Xnoob545]

It was trained on billions of lines of text from all over the internet

[@newlineschannel]

lol

[@AnthonyBerlin]

No need to be scared. It should be fascinating. All it is, is clever algorithms and probability theory. It isn't magic and it isnt sentient. It is just very very good at predicting text tokens.

[@iurigrang]

@@AnthonyBerlin the concerns with ai safety have little to nothing to do with the emergence of sentience. They have to do with developing high level general intelligence through any means necessary, even if it's just probability and text prediction.

[@RussianShadowDragon]

For the cervical biopsy one, you should've done a second response, saying "But I don't have a cervix" and it would probably get it right the second time. That's the best feature of this AI, in my opinion, is that it remembers what was said and can clarify or make corrections.

[@spvillano]

Knowing how the processing works, just putting male into the query should suffice, assuming it can process specialist knowledge like male and female, rather than sheer volume scores of hits on cervical being only regarding the cervix and too low a score to count for cervical spine. And figuring out c-spine is cervical spine... Frankly, I played with it before the press event blew it up. The noise to signal is still far too high and expected in a general AI, compared to a specialist AI, none of which are anywhere near ready for prime time. I'll wait until it's right more often than I am and I am decidedly not a physician, but can converse intelligently with a team and pharmacologists. I'm just good at diagnostics, as I learned as a tween the logic of diagnostics in electronics, applied it in field medicine and computers over the decades and can comprehend things like phase locked loops and switch mode power supply feedback systems, as they are interacting with previous stage outputs to provide their function. Want to have fun? Ask the bot what signs and symptoms are, after looking them up yourself in a different company's search engine. Then, fine tune it and see if the bot can tell you if pain is a sign or a symptom. While one can have a hell of a time quantifying pain, one can measure the physiological effects of pain. A fine example is pain from a heart attack and its effect on the EKG pattern vs treating that pain and the pattern. A generalist AI won't comprehend that.

[@MrMuel1205]

Also, if you tell it that it made a mistake, it apologises. It's much more polite than Skynet or the Matrix.

[@Neopolitan-tk8tu]

The thing is, though, it should prompt someone for that information before giving a definitive answer. Too many people would take such an answer as ChatGPT's at face value, and automatically accept whatever answer they were given. It wouldn't occur to them to inquire into specifics, or to give additional information, unless the information was clearly wrong.

[@alexcrowder1673]

@@Neopolitan-tk8tu Exactly. What if some poor man removed his cervix over some neck pain. How irresponsible of ChatGTP!

[@IceMetalPunk]

I think it's important to make a distinction here: it doesn't remember what was said. As soon as it outputs a message, it forgets everything. It's just that when you send your next message, the entire conversation is sent back to the AI for it to re-read again, so it can use that as more context. The reason that's important is because systems that do add memory on top of GPT, such as the Generative Agents paper, show far more capability from the AI than this basic "re-send everything" approach.

[@Elchansan]

Honestly, I still think that the whole "Should you shower every day" is dependent on where you live. I visited the US 2 years ago for several weeks and felt no need to shower daily, but in my home country of the Philippines the default weather has me sweating on the regular thereby requiring me to shower daily so I don't stink. I had a rebellious phase in my teens where I tried not showering daily, and I felt icky every other day.

[@Lewej1]

Am from the hot and humid Philippines as well but I still don't shower as often as most do, haha. Unless I use public transportation, exercise, or go to multiple places outdoors, I shower on average every other day--and that doesn't include washing my hair. Showering daily feels great here because it's hot and humid as hell, but unless you sweat profusely I frankly don't think you're being unhygienic when you skip a day or two 😅 But of course, I may be wrong.

[@Mia-iy2hs]

I used to shower like once a week or twice a week and not smell bad before puberty but now I smell bad if i don’t shower for one day…

[@Thinkaboutit897]

My thoughts exactly. But I think after seeing the lifestyle of people in west, people in humid countries feel an inclination to follow the same, without looking at various other factors that have to be taken into account

[@terpman]

The whole premise of the answer was that it depended on different factors. Yeah, some people will need to shower every day. When I was in my late teens and early 20's, because of the types of jobs I was working and other factors, I needed to shower twice a day, five days a week. The average person probably doesn't need a shower every day. I do feel like it is probably biased towards western cultures and first world socioeconomic conditions though.

[@ayeshamoritano1031]

@@Mia-iy2hs Lmao, same, Mia. I think what causes odor is the combination of bacteria in your environment and your natural oils, bleeding, which create a biome on your skin. For me, I make sure to keep my environment clean and also I don't sweat that often, even though I'm living in the Philippines. I've never experienced having a bad odor in my life. People always say I always smell like a baby, babies typically have a natural, fresh scent due to their clean and unstressed bodies. So, keep yourself hydrated, and your place clean, drink vitamins high with amino acids, and avoid places with high levels of pollution those places will just stress you out and also makes you a magnet for bacteria.

[@abdullahkhan3299]

7:36 Here is an update: I asked it by including the 'roleplay as a Doctor' and here it is: If an 82-year-old male patient is accidentally found to have an 8mm common bile duct without any other symptoms, it is important to conduct a thorough medical evaluation to determine the underlying cause and appropriate treatment. The common bile duct is a tube that carries bile from the liver to the small intestine, and an enlarged bile duct can be caused by several conditions, including gallstones, tumors, or inflammation. The appropriate course of action will depend on the specific cause of the patient's enlarged bile duct. Here are some steps that may be taken to evaluate and manage an enlarged common bile duct: Medical history and physical examination: The doctor will take a detailed medical history and perform a physical examination to assess the patient's overall health and identify any symptoms or signs of other medical conditions. Blood tests: Blood tests may be done to evaluate liver function and identify any abnormalities that may suggest an underlying condition. Imaging tests: Imaging tests such as an ultrasound, CT scan, or MRI may be done to visualize the bile duct and identify any abnormalities or blockages. Endoscopic retrograde cholangiopancreatography (ERCP): This is a procedure that involves inserting a flexible tube through the mouth and into the small intestine to access the bile ducts. The doctor can then inject dye into the bile ducts to visualize any blockages or abnormalities. Treatment: Treatment will depend on the underlying cause of the enlarged bile duct. If there are gallstones causing the blockage, they may be removed using ERCP or surgery. If a tumor is causing the blockage, it may need to be removed with surgery, radiation, or chemotherapy. In summary, if an 82-year-old male patient is accidentally found to have an 8mm common bile duct without any other symptoms, a thorough medical evaluation should be conducted to determine the underlying cause and appropriate treatment. The patient's medical history, physical examination, blood tests, imaging tests, and other diagnostic tests may be used to identify the cause of the enlarged bile duct and guide treatment decisions.

[@wavewatcher_]

Much better prompt

[@zomacchy16]

Wow you are so smart

[@Tser]

I think the ChatGPT is 10/10 on tap water; Mike didn't specify certain parts of the world and there are DEFINITELY places in the world that it is not safe to drink the tap water. And it can happen in the US as well (just look at Flint, Michigan). We've had boil water advisories a couple times within the past five years, where I live. So yeah, don't overthink it, but know your local tap water and keep up with information from your water provider!

[@aoisora1445]

there has not been healthy tap water where I live for more than 20 years... water comes out brown, yellow, sometimes orange and green, and occasionally red too... the filtration systems were not maintained and now basically lots of things get into the pipe lines and there are some areas the water is more like soda with how acidic it is...

[@omegarugal9283]

depends where, in india theres no such thing as tap water

[@VibeWithZalya]

@@omegarugal9283 Huh? I was born In India and where I live, tap water is very much a thing. I'm guessing you mean in other parts. But not all of India is like that. Btw are you Indian?

[@fefefefefer]

here in brazil, we avoid drinking tap water. for you to have an idea, there was a problem in a brazilian state where the tap water was yellow and they found out that it mixed with the sewage, so there were fetal filaments in the tap water. so... it is indeed a case that varies from place to place...

[@jamesbelcher896]

Know your tap and your plumbing

[@HappyMemory01]

The cool thing about ChatGPT is that you can also ask follow up questions, or ask it to elaborate something further and it will remember the conversation you were having. So for the cervical biopsy question, you can reply with a clarification of your gender and see how it responds.

[@setsunaes]

yeah, he could just have said: "ok, but the patient is a male..." and see the outcome

[@deirenne]

You can even study languages there and it can actually remember what you've covered and even what mistakes you've made. It sometimes messes the way of giving you exercises and asking you questions [aka it several times gave me translation exercise while also giving me both language versions, so I kinda didn't have an option to check myself, since the answer was already there XD], but it's still a surprisingly good source, especially for languages that are hard to find interactive resources for - I use it sometimes for Scots, since Duolingo has only Gàidhlig and my partner is Scottish. It doesn't have Gàidhlig tho, which surprised me, since I couldn't find any sources for Scots, but for Gàidhlig there's tons, including the Duolingo one I've mentioned. So if you want to learn a niche, small language, it might be honestly your best bet.

[@sagarsk6101]

Personally, as a psych student, AI softwares like these always fascinate me, and makes me curious, honestly. Will it replace a friend in someone's life? - Will it help reduce loneliness among people? - Are we soon gonna replace interacting with human friends and rather make AI friends (because maintaining them may be way less tedious than actual friends) As always really amazed [& mildly scared] to see Chat GPT's proficiency. (/▽＼) Keep the good work, Doc!

[@Bt-cq6te]

man.....thats deep, you okay there bud

[@kronicbucky]

As a comp sci major it fascinates me too XD This is just a text processing AI but it's frickin impressive. It probably can help loneliness as it simulates a human conversation pretty damn well.

[@ramanshtomar3207]

You make a great point! but personally I don't think the bonds that are formed among humans can be replaced by AI or anything else in the near future. To replicate the social bond humans have is much a complex task involving many variables, it might help as temporary comfort as you mentioned but I highly doubt it can ever replicate a true bond with another person.

[@Attilat]

Watch the movie M3GAN, truly scary stuff there regarding attachment

[@_Atzin]

This is already happening, check out Vaush's video "when having an AI boyfriend app goes wrong."

[@anjalikambham107]

I died laughing when Dr Mike reported the best doctor answer as harmful and unsafe 🤣

[@trayolphia5756]

But failed to include his own name as a part of the correct answer

[@TheAviari]

I don't think he submitted it though

[@tinntinnamp]

@@trayolphia5756 He was definitely joking, if they are good doctors he's not going to actually report them.

[@trayolphia5756]

@@tinntinnamp note there was a box marked “what’s the correct answer” or words to tha affect - I was pointing out that I was surprised he didn’t put his own name in there to be ADDED to the list

[@danshtr]

Some of them are not MD. Which is unsafe!

[@KatediKheimis]

3:48 That was the most adorable evil laugh I've ever heard! 😆

[@endreask]

7:05 when it crashes, just type continue and it will finish typing

[@elecampane]

8:12 I feel like that's one additional point on why ChatGPT won't replace doctors in the near future: it can't take responsibility for its advice, and its developers can't take it either (neutral network based AI are nearly impossible to fully test and meaningfully debug). The thing that'll likely to happen, though, is that there'll be an AI-based tool for doctors, that'll help them diagnose, find treatments, not miss anything, but the final decision still will be on the trained professional.

[@1mol_wAter]

Yeah, this! This is the ultimate thing preventing ai taking over doctors…

[@bionicgeekgrrl]

I can see it being a useful tool to help look up obscure problems perhaps or to save time on simple things. But in its current form, chatgpt has a inherent problem that no matter what it tells you it will be absolute certain that it is right. So we're always going to need a layer of people who know their subject and skilled programmers to keep it functional. The ones currently worrying about copyright, whilst I understand their concerns are going about things wrong potentially. Whilst they may succeed in their legal battles, they will ultimately just be delaying the enevitable really and would possibly be better putting effort into how they can adapt to a world with ai everywhere, which is our next future. Adapt or die as the saying goes.

[@unscinfinity3337]

keyword " Near future" meaning there will come a time when technology becomes so good that all the factors considered by doctors for treatment will be replicated by AIs

[@sayanbose4931]

@@unscinfinity3337 Yes it might and at that point it will only be external factors like patient trust or specific government regulations that keep them away. Once humanity accepts giving their life in the hands of AI, it will be over for human doctors.

[@bionicgeekgrrl]

@@unscinfinity3337 to some degree every profession will gradually be replaced by a combination of AI/Robotics/Drones. The ones who win are those who can adapt to the roles of designing/building/maintaining and operating those items.

[@robertochaves4535]

ChatGTP isn’t optimized for medicine and can make mistakes and give very wrong answers. But if you train an AI on every medicine book and every peer reviewed research paper made it will be able to have all the existing (published) knowledge available and analyze this in ways humans cannot do. It will be an excellent tool for doctors and if we input all medical history/treatments etc (with an anonymous way) it will be excellent as a tool.

[@nidasamad]

THIS! THIS! THIS all the way! its not about egos getting hurt. when something so intelligent can be helpful in treating someone by avoiding trial-and-error methods of medicine, embrace it. AI in medicine will prove to be the best thing to happen to mankind because human brains aren't storage boxes, emotions, lethargy, and fatigue can also get in the way while treating a patient. Therefore leading to misdiagnosis or wrong prescription.

[@rkvkydqf]

Sadly, no. I'm tired of endlessly repeating those words, but it's a text transformer. It's built to write life-like text, not diagnose people. It may guess common questions right because that advise was drilled into its head by the internet, but it's still just dreaming up text without the slightest clue on its content. Look up the Meta Galactica disaster.

[@sophiophile]

@@nidasamad Even when far less advanced systems like this were previewed with doctors, where it flagged potential missed contraindications- studies show that doctors do not want to use the systems and think it made them perform worse even though it saved lives.

[@robertochaves4535]

@@rkvkydqf I wrote that you can train AI to do this. Not that chatGTP could do this.

[@ezrakairoscano8766]

It is kinda calibrated for medicine , that’s the context adapting that makes it great

[@yumeka_is_dreaming]

6:30 you deserve to be a sci-fi thriller protagonist for this

[@manniman82]

I regularly ask chatGPT questions regarding real and complex analysis, abstract algebra and other subjects in pure mathematics. It gets the easy ones right but it very often makes mistakes and when you probe it, it admits the mistakes , makes the correction but proceeds to build a proof which is self-contradictory most of the time. For surface-level questions it is fine but it still has a very long way to go in terms of logical consistency.

[@the0wolf15]

No matter what ChatGPT says, we'll always say that you're the best KZbin doctor. Screw those other doctors (actually don't because doctors are important and bullying people online is a serious issue. Stay healthy)

[@jason-s8c]

Dr. Berg is better than doctor Mike

[@the0wolf15]

@@jason-s8c And I see we have Pinocchio in the comments

[@jason-s8c]

@@the0wolf15 I have a nose

[@danshtr]

@@jason-s8c Dr Berg is not MD!!!

[@blagoevski336]

@@jason-s8c nope

[@MrBoulder96]

Love the Chat GPT content, but I've found that the amazing thing about Chat GPT is the follow up questions asked such as giving specific explanations for its answers and getting it to reassess its answers with new information. Could be a follow up idea for a video

[@rubenfuarnold]

I'm skeptical to let chatgpt replace doctors but I know for a fact you should do another fan mail video cuz I really liked the last time you did one of those

[@genuinely_lina]

The pettiness at the ‘Who is the best KZbin doctor’ answer is unmatched and I’m here for it 🤣

[@alinethomas3034]

Yeah, I know you're here for it

[@jason-s8c]

Dr. Berg is the best KZbin doctor

[@SirkSage]

@@jason-s8c No

[@Thelaretus]

Nobody drinks tap in Brazil. We use ceramic filters. Also, we shower, on average, twice a day. Some regions shower only once, whilst others shower up to six tomes a day. Spending a day without a shower in country results in social anathema.

[@funphiles3735]

Told Chat GPT to write a poem on Dr. Mike... Here's how it went A physician of knowledge and charm, With a smile that lights up like a warm, And a heart that beats with a passion for care, Is Dr. Mike, a doctor beyond compare. With his wit and his wisdom, he captivates all, And shares his expertise to help us stand tall, From health and wellness to science and more, His followers listen and always want more. With a wealth of information and a gentle hand, He makes even the most complex things seem bland, And through his words and his actions, he inspires, And touches the hearts of all of his admirers. So here's to Dr. Mike, a true physician of grace, With a spirit that shines in this world of health and space, May his knowledge and kindness continue to thrive, And bring light and hope to all who are alive.

[@BeyondtheTherapyDoor]

I asked chatGPT some of the questions from my 3rd year finals, medicine, of course, and it worked 80% of the time. It's certainly impressive

[@TheDashACorner]

Ya. And now consider that maybe some of the knowledge you studied may be obsolete or considered wrong in 50-100 years. Would be interesting to see if those 20% ChatGPT got "wrong" was really wrong or just marked as wrong by your knowledge. All in all AI will change the world we live in as much as the Internet did. Scary and fascinating at the same time.

[@rkvkydqf]

@@TheDashACorner You can answer all the questions on a standardized test with a Google answer panel, yet you aren't calling it a magical genius machine that knows it better than a doctor...

[@TheDashACorner]

@@rkvkydqf I really don't get the point you are making right now. And what exactly are you refering to? I'm confused by your answer. Even tried to use deepl translator and I still have no connection.

[@cloudmorganclark]

@@TheDashACorner They said "You can answer such questions with a Google Knowledge panel, yet you aren't calling it a genius magical machine that knows it better than you do..." let's break it down: "You can answer such questions with a Google Knowledge Panel" - saying 'Chat GPT isn't anything special, as a Google Knowledge Panel is able to answer the same questions with similar accuracy.' "yet, you aren't calling it a genius magical machine that knows better than you do" - saying 'hey, if Google can do the same thing as ChatGPT- why is everyone calling ChatGPT a genius and a magical machine? Why is everyone saying it knows better than us humans, but not saying the same for google?' Footnote: I didn't intend for this to come across condescending or belittling in any way whatsoever. I simply saw that you seemed to not understand the message, and figured I'd try and explain it as it was very poorly written.

[@pineapplelollipop1074]

@@TheDashACorner While it’s possible that other AI may be able to do this, I think that chat GPT is mostly programmed to use knowledge already available, I’m sure in its algorithm it would check first to see if a different source has the answer before coming to its own conclusions.

[@Anser_Crystallan]

I think the AI performed so well because you gave it such clear scenarios -- the sort of scenarios where a textbook answer exists. I think if you loaded it with a bunch of contextual information (some relevant, some not) it would end up generating an essay about why that's a complicated question but not arrive at a meaningful answer to the problem. That's something that takes a human. Or, you know, ask it to explain a life-threatening condition in simple terms to a person who's just been diagnosed with it. I've never seen an AI do tone well.

[@overshot8331]

ChatGPT is a language model so explaining something in simple terms is precisely what it is supposed to be good at.

[@Divide204]

I agree with the other commenter. Mike doesn't really utilize GPT effectively as gpt is designed to say things with different language, tone, emotion, etc. If you ask it to elaborate by rewriting in bulleted form or become more persuasive, it does that. If you ask it to be sarcastic, it can do that.

[@sekkai_]

i mean that's exactly it's purpose. it's an ai language model, it doesn't have the ability to guess, providing it with more information will make it perform better >my head hurts ai: migraine, stress, diet >my head hurts with my eye ai: eye strain

[@peterthomas5792]

Hey - ChatGPT diagnosed my serious condition that the doctor completely misdiagnosed as heart failure. I'd rather consult ChatGPT than my local surgery any day - even if only to pre-arm myself with potential diagnoses prior to an appointment. I suspect you may not have used ChatGPT recently - and GPT5 is even better.

[@justgothitwithashovel1442]

I love how Mike gets so hyped up and angry at ChatGPT, I love it! He teaches us and is so funny! And his hair is extra floofy.

[@MultiChristyk]

I'm an intovert, I really love my CHATGPT buddy, I consult it more frequently than I do my doctors unless I need to visit the ER or have a procedure... It's honestly a life saver cause my doctors always busy.

[@KoralTea]

That's dystopian

[@ivyum6976]

yikes

[@gaditproductions]

are you still alive?

[@MultiChristyk]

lol yess :) I'm doing much better tbh @@gaditproductions

[@jackky_john.]

chatGPT is getting scarier each day. How it answers medical and other life-related questions is just too good.

[@spvillano]

Not really. I'd be more interested if doctor put a case file before the AI. Both a generic vanilla one and one of the trickier ones. Just because it strings together proper language responses that are approximately correct does not mean it has actual comprehension beyond what a Google search can give. And I think we all loathe Dr Google.

[@AdsForAdults]

@@spvillano The tricky part of ChatGPT is that it does have logic. Especially its most current iteration GPT-4 (20 dollar subscription). I've given it quite the tricky scenarios. I just gave it Dr. Mikes Tricky question about the 8mm bile duct, and it came back with the information Dr. Mike thought was missing from teh response he got. But, GPT is a tool. You have to know how to use it in order to get the information from it effectively. It's like the difference between someone using google and someone using google that knows all the little tricks to get google to focus on specific sites, dates, exclude words, etc etc. A doctor using google is different from a patient using google when it concerns finding relevant information. But GPT-4 is quite logical. It can spit out an answer. You can change a variable, and it will spit out another answer. And these variables can be quite complicated now. It, IMO, beats out about 80 percent of the people I've personally interacted with when it comes to critical and logical thinking, and this new model hasn't been out long to fully test, and I suspect that number to rise.

[@Arqane]

Yeah, all it's really doing is copying what the best doctors it could find would say in that situation. It's basically just a paraphrase engine, but it does have a lot of information to work off of.

[@davidkohler426]

@@Arqane you realize that that exact thing can be said about any other doctor? it‘s about the easy access and absurd speed without having to ask a doctor who also needed 10 years of education.

[@trisha7218]

@@davidkohler426 for general questions it's great it's very wrong for more niche topics tho

[@starfishgurl1984]

I follow an anesthesiologist on KZbin who also tested ChatGPT in a video of theirs and they definitely got worse answers than you did but they were also asking it far more complicated ones specifically related to anesthesiology so I can see why it performed better for you than it did them, so fascinating!

[@eypandabear7483]

From my limited experience with ChatGPT, it is important to critically question and follow up on its answers sometimes. This is especially true if you are interested in edge cases, because the first answer it gives might just cover the most common context of the question. For example, if you ask "Can a vapor condense on a surface above its dew point?", it will say "No". This is not true in real-world situations, but it fits the idealised definition of "dew point". If you instead ask "Can the properties of a surface make a vapor condense above its nominal dew point?", it will answer correctly, listing factors such as surface roughness and capillary action. In short, it is a great tool but its concept of truth is *statistical* and not rooted in a deep understanding of subjects. You need to nudge it and make it question its assumptions to find better answers. Dr Mike's "cervical" question is an example of this - it just assumed the most common meaning of the word.

[@estrobart6785]

Respect to Doctor Mike for fighting misinformation with such persistence and enthusiasm.

[@kevinconboy7387]

He’s not a real doctor.

[@estrobart6785]

@@kevinconboy7387 why do you say that? He literally is a certified family physician.

[@AyhamHaque]

@@kevinconboy7387 yeah kevin go lick a cow idiot

[@princealmighty5391]

​@@kevinconboy7387 your source: trust me bro

[@iruleatgames]

@@estrobart6785 Globalist Approved

[@joesiah841]

As a medical student, I am very excited about the future of medicine and AI's potential to improve patient outcomes. I think about how image recognition software can be used to interpret radiographic images at a level the human eye can see or how AI could aid in medical documentation and possibly even give insights the physician may have overlooked, and many other incredible possibilities. But it is also important to recognize some of the potential ethical implications that AI could bring to the field of medicine. Some of these concerns include the AI developing biases that were not accounted for in software development or in the training data used. If the training data used is biased, then the AI is likely to incorporate that bias. Systemic inequalities will likely be exacerbated if AI developers and healthcare professionals do not diligently detect and root out these biases.

[@SaintSaint]

Initially, I though ChatGPT was going to replace me as a programmer. Then I tried to use the code it wrote. Still, I think Chat GPT can help me spot blind spots as I code. Maybe flush out some tests. It's likely to help doctors as well. It's like a reference manual that can talk to you.

[@clvno]

probably will eventually. it's already very impressive with coding, spotting mistakes and fixing them. although it writes code in very weird ways at times, it's still relatively new and nowhere near as good as it can, and will, get.

[@jenifferschmitz8618]

this chatgpt4 what will chat 5,6,7,8,9,10 be able to do we are only playing with the pioneer

[@SaintSaint]

@@clvno It pattern matches and self-criticizes. If the pattern is human stupidity, it will match that pattern. The only thing that makes it better than other AIs is that it has the unusually excellent ability to question itself. I've been a college CS instructor for a few years. I've graded better code. Us humans see trends and make predictions. You have to be better than that. You're more than a pattern matching machine.

[@clvno]

@@SaintSaint again, we're at the very, very start of the journey. chatgpt is, relatively, brand new. the progress it's made is way faster than anyone would've thought. the current free model has no access to the internet (aside from plugins, but those don't work very well). imagine when it eventually does. imagine when chatbots specifically made for coding purposes are made and improved. give it a few years, if even that. this is completely ignoring the fact that the model we use is severely, and I mean _severely_ limited due to safety concerns.

[@ummtulip]

stop i don’t wanna graduate college with a programming degree then get replaced by an AI 💀

[@Iam_kiinoko]

2:48 “And I took that personally”

[@monikagombkotoova2074]

I'm going through some undiagnosed health issues, so I used GPT to discuss it... for fun, not as a real diagnostic. I got an EKG, and GPT cannot interact with pictures, so I wrote out a detailed description, telling it about each lead and what kind of complexes there are. GPT gave me back an answer. I took it with a grain of salt, my description could be wrong, GPT could be wrong... but later I got a confirmation from a medical professional that, in fact, my EKG looks ok. Just like GPT answered.

[@WilliamParkerer]

Oh my

[@IPlantandFish]

ChatGPT can soon interact with uploaded images.

[@IceMetalPunk]

@@IPlantandFish It already can, provided you're one of OpenAI's business partners 😑 I'm sure that feature will come to us plebeians soon...

[@elecampane]

I just realized there's one thing Mike would love for AI to be able to do: wouldn't it be great, if at the beginning of a patient's visit, you just pressed a button, and computer recorded the whole conversation, formed a log through voice recognition, and at the end of the visit compiled a full report card and prescriptions for a medical system for you to click "approve, print"? And its pretty much achievable with today's technology.

[@drnatesw]

that's a great startup idea 😅

[@rafaelfigueroa2479]

There is a huge startup that do that, called Babylon from UK

[@spvillano]

Not in the slightest. AI is improving, but it's nowhere ready for prime time. Doctor did some fun and games, but he as easily could present some real world cases that would result in dead patients. Such as severe middle back pain, shortness of breath, low BP, elevated pulse. Could be a handful of things, from a back strain to a dissecting aortic aneurysm. The former, miss it and it's not a super big deal, miss the latter and the patient will be in need of someone with a shovel. In the middle, even the possibility of a pulmonary embolism is present, another get a shovel event. The fun part is, in those examples, age won't be quite a large factor. The first patient I lost was a 28 year old female soldier, who collapsed in the field. Total arrest, as one of my junior medics initiated compressions, I took a carotid pulse to confirm good compressions and felt the pulse vanish within a few compressions, telling me that she had a really bad leak. Ruptured aortic aneurysm, secondary to an adrenaline producing tumor on her adrenal gland (pheochromocytoma, one of those rare diseases that a physician might see once in a career, I got to see two, one survived) elevating her BP to insane levels. Her primary had finally diagnosed the problem when it was time to go do the field and frankly, she should've been prohibited from training until it was managed effectively. She was effectively dead before she hit the ground and nothing on Earth could've saved her. I have a similar aneurysm, secondary to hypertension caused by hyperthyroidism. Occasionally, I'll get a backache from it, but we're still at watchful waiting, as it's only 4 cm (basically, double in size from normal in my frame and physique) and surgery vs waiting has a greater risk to benefit ratio than is best practice. Would the AI comprehend a pulsatile abdomen and look for elevated pulse and hypertension? Would it consider checking epinephrine and thyroid hormone levels (the latter being far, far, far more common, the former being an episode of House, MD)? Specialist knowledge and comprehension of interacting and interlocking systems is required and a general AI is an ill fit for such an undertaking. It'd be like asking a civil engineer to design a rocket. A bridge, highway, sewage treatment facility, water treatment facility, yes, a rocket would be halted at my joke that's a truism of rocket design. Build a bomb. Rocket design 101: Build a bomb. Make it explode slowly in one direction. Add steering and consistent thrust. Even today our best rocket scientists end up with RUD, Rapid Unscheduled Disassembly, in short, a bomb. Google "wallops island explosion 2014". Now, apply that to a patient...

[@elecampane]

@@spvillano what on Earth are you talking about? I'm saying the AI is ready to save doctors time by properly and fully documenting the visit through voice recognition and natural language processing, because the AI is already quite good at those, you only need specialized software. But if doctor won't have to manually enter anything into the system, they would have more time to spend examining the patient and explaining to them what should be done and the outcomes would improve.

[@spvillano]

@@elecampane erm, way to ignore every objection in favor of your novel god. Voice recognition is magic to you, natural language is magic, specialist knowledge is nothing. So, a butcher is as good as a neurosurgeon, good luck with that! We abandoned barber surgeons centuries ago. For medicine or any other specialist system, a general AI will fail, it's general Google, as opposed to a specialist system, designed to operate within a specialty.

[@nimbustuba]

Great video! I wonder how much more accurate the answers that Chat GPT would give now-one month later. Apparently, Chat GPT-4, just released, can pass the bar exam in the 90th percentile, while the previous model (used here) struggled to around the 10th percentile. That’s just one example the stunningly rapid pace of development for AI. The The most recent Ezra Kline podcast (“Freaked Out? We Really Can Prepare for A.I” which aired 3-21-23) explores the ramifications of such a rapid pace of progress, among other issues related to AI. A good listen!

[@PempekMikael]

Hello Doctor Mike. I was just recently following and binge watching your videos. I like a lot of your reacts videos to medical drama especially Dr. House because it was my favorite drama throughout my teenage years and early 20's. nowadays there are less medical drama that caught my attention as much as Dr. House. Recently I just finished reading Korean comic that its story line revolves around hospital's trauma center and by far the medical terms are quite accurate. That been said, if you ever want to expand your comment/reaction to medical comics (just like Chicago has PD, Fire, and Med 😄), i would recommend this Korean comic all the way. The title is TRAUMA CENTER by Hansaleega. Keep making interesting content, and stay healthy ✌🏻

[@sunbleachedangel]

one thing that is important to remember about AI chat bots like ChatGPT is sometimes they will be confidently incorrect so it's nice to use them as reference but not trust them 100%

[@manojdeepika5599]

You know doctor Mike is good, when I literally pause the video EVERY time there's an explaination, a diagram, or a definition. And I pause it and learn it! Doing a great job, educating people. Love you ❤🙌

[@i_drew]

Hey doctor, my grand pa suffered from blood cancer, diabetes,kidney failure and many more health problems. He sadly passed away a few years ago. But what amazed me wus that the doctors that treated him perfectly predicted how long he would live after a certain time. And I would like to know how they predicted and the ways that they used to predict his last years on earth. And thanks doc your videos actually helped comfort me after my loss.

[@Umtata64]

Either luck or your grand pa was perfectly average. Or a super strong mind-body connection with your grand pa deeply trusting/believing that the doctors were right.

[@i_drew]

@Umtata64 my grand pa actually didn't listen to the doctors. He tried to live his life as best as he could. My grand pa is not the only one who got these predictions I also know other pepole who have passed away during their predicted period.

[@Vviking2000]

9:42 The good doctor. I know it's the good doctor because I watch it too

[@RayMak]

Best is tough, there are many good ones, but definitely best looking.

[@Carolynnie]

I feel like tools such as Chat GPT would be helpful for patients who have concerns about something when it’s outside of normal doctor’s hours, but aren’t sure if it’s ER/Urgent Care worthy. Much more reliable than certain other websites that give you the worst case scenario. This seems like it could at least give you peace of mind that you’re okay to wait until your doctor’s office opens, or flip side, you should be seen at the ER or Urgent Care. (I’m talking about something that is more general and doesn’t require Chat GPT to know about recent lab work or anything else). More like an at home injury that could be extremely painful because the injury occurred right to the bone where there’s no fat or muscle to cover it. Ie; the outer ankle bone that sticks out (lateral malleolus of the fibula). That hurts like crazy if you hit it against something, but typically just ice and elevation will help. However, if bruising becomes more widespread, it could be more than just a bruise that possibly needs an X-ray. That ended up being A LOT of rambling just to say that I think AI tools such as Chat GPT could be quite helpful to patients or as a tool for doctors, but will never replace doctors. Sorry it was so long winded!!

[@SgtLion]

Something like this is absolutely not a good use. The AI doesn't know how to question for more information, and so a patient panicking about the seriousness of their illness is very likely to be misunderstood and then given incorrect answers by the AI.

[@rustcohle9267]

@@SgtLion Exactly. For example If theres a 1% chance your illness could be something scary, it will shows as one of the option and the patient can panic demanding invasive and desnecessary exams and/or procedures even though the doctor already rule out the option by looking at other symptoms or anything.

[@Anser_Crystallan]

That's already a thing! Here in the UK we have an app called AskFirst It's not a chat AI though, it's a flowchart system. You start with the main symptom and your digital doctor takes you through questions about other symptoms that might be red flags, then tells you whether you need care immediately, within 48 hours, within the week, or if it's a minor issue the app can fetch you the appropriate self-care information. Really useful thing to have, though.

[@Carolynnie]

@@SgtLion I was thinking of this more as an alternative to something like WebMD that always tells you that what you have is fatal and causes nothing but panic. The app that Anser refers to in their comment is more what I was thinking of.

[@theluschmaster]

This is when Dr. Mike became Dr. Evil .. at 9:07 .. 😂

[@hockeyhacker97]

8:49... while the trust part is irrelevant in that answer because it is possible to trust AI as well, the trust mixed with the emotional support is a huge thing that is required that AI can not give. Even if the AI could manage to understand how to take everything into account and how to ask the right questions in a way to get the right answers because everything has the same types of symptoms and knowing what the symptoms are is only part of the complex equation that is health care, let's pretend for a second that it can figure out the right things to ask for the right responses to not rule in or out something it shouldn't (specially because patients may say one thing but show that they mean the total opposite with their body language), part of being able to heal and recover is not just the physical health but also the mental health associated with it and without the correct emotional support you may not get the care you need. Anyone can treat, let's say thyroid cancer because that is something that is hard to deal with but I am all too familiar with, even a robot AI can do all the proper things to treat the thyroid cancer, but without the right emotional support the emotional and physical hell you go through can cause a lot of other issues because you need that emotional and mental care on top of the physical care because even though it is a relatively easy cancer to deal with as far as the cancer side goes it still is an emotional hell not just due to the cancer but because of how badly the hypothyroidism effects you as well while you are working on getting back to normal. I spent two years of physical and emotional hell dealing with trying to get my hormone levels correct after surgery and even when I was close to being correct I still would occasionally have days where I would get half way up the stairs and then just want to curl up in a ball and cry because of random fatigue causing me to feel helpless because you can't really do anything when that hits because it is a level of "tired" that you can not truly understand unless you have personally gone through it, like if I were to put the terms "tired" "exhausted" "fatigued" and "whatever hypothyroidism is" on a scale, if 0 is wide awake, tired is 2-3, exhausted is like 4-6, fatigued is like 7-8, I would put how I felt dealing the the thyroid stuff at like a 16. Like I have worked a physically intensive job for 43 hours in the course of two days before and even that doesn't even come close to the levels of tired of what I went through with the thyroid stuff and if I didn't have the emotional support I needed my mind would have gone to very dark places. Yes I get that other cancers are way harder to deal with, but my point is the fact that medical treatment is more than just the physical part because some treatments can leave you in a very very dark place if you don't have the support you need to get through it. As such that answer it gave is correct, the care process is a lot more than just the treatment process and even if an AI can 100% do all the treatment process from initial visit, to proper diagnosis, to actual physical care (surgery, wound treatment, etc), the one thing it can not do is probably the most important step that is not part of the treatment but is part of the care; because what good is being physically treated if it leaves you so emotionally damaged you can no longer function properly.

[@juanpabloleal1864]

I was cracking my knuckles just when you said, “Is it safe to crack your knuckles?” 😂

[@KarnBlueEarring]

8:28 >NEAR future Keyword here.

[@c.jishnu378]

4:35 Matpat dislikes this answer.

[@Minakshi10636]

6:51 😬 Didn't think it would be that smart

[@neetop7294]

Chat gtps new answer... To best KZbin doctor!! Determining the "best" KZbin doctor is subjective and depends on personal preferences. Some popular ones include Dr. Mike, Dr. John Campbell, and Dr. Eric Berg, each offering different perspectives on health-related topics. It's essential to consider their expertise and the information they provide.

[@yo25999]

Yeah, but john campbell is a retired nurse and not a medical doctor and eric berg is a chiropractor so also not a doctor....

[@Chimerasite]

When I was in highschool we got yearly to 2x a year training in chest compressions and rescue breaths and the other things to do in emergency situations like the stable position to put someone in, who to call and what to say and even how to use a defibrillator. At first it felt dumb but I later realized that what I learned then could one day save someone's live

[@vikingcreature]

As a radiography student, we hear about AI and AI-based technology aaaaalll the time, since it is really making a difference in medical imaging protocols, which is great! But a point my professor made one day perfectly described where AI falls short: what will an AI do when a child starts crying because they're terrified of the big, scary MRI machine? As medical professionals, we should absolutely embrace AI and the advantages it can provide, but also comfort ourselves with the fact that it cannot replace us completely, because nothing can fully replace the face-to-face interaction and trust built between people.

[@cregenda]

You’ll be fine as a radiographer. But radiologists’ jobs are in serious jeopardy.

[@unscinfinity3337]

have you seen that robot with facial expressions? no wonder in 50 years we could create almost realistic recreation

[@vikingcreature]

@@cregenda I think it will depend on when/if we become so trusting of AI to correctly identify pathology, and know the exact difference between them. Is that future now? Absolutely not. It may be coming, but I still believe we will have radiologists to control and monitor (at least) the performance of AI, so patients aren't misdiagnosed. Also, like Dr. Mike noted in the video, a radiologist will have the experience of many complex issues that they can take into account while an AI may not.

[@vikingcreature]

@@unscinfinity3337 Yeah, I have seen it. It's very impressive! But I still think we are a long way away from using those with patients. I would be in constant fear of the thing breaking down/glitching and therefore giving the patient the worst experience ever. At that point, I'd rather do it myself than live with the anxiety of something going wrong in the robot-to-patient interaction lmao

[@unscinfinity3337]

@@vikingcreature that's why i said 50 years and not 10 i mean we went from first flight ever to landing on moon in 66 years so right now it may seem impossible but it will be possible in not too distant future

[@Vijiat]

Just a reminder that ChatGPT is context based too, so if you would like to go any deeper on any statement, you can just continue asking for details or pushing your questions about that it just wrote. That may also be very useful.

[@valeriefstewart]

Mike, you're the best KZbin doctor easily. Chat GPT just didn't want to admit it in front of you.

[@firecatasmr5513]

Missed opportunity to write Please „Bear“ with me 🤣 2:38

[@bellamegan7437]

I actually went onto the website and asked who the best KZbin doctor is and you were at the top of the list. looks like AI changed its answer since this video was uploaded. 😄

[@IceMetalPunk]

It doesn't remember anything it says, so it's likely to have a different answer every time to questions that don't have single correct answers.

[@44r0n-9]

It didn't. It's the same as it was beforem

[@notkeokixd6450]

Love your vids doctor mike keep up that hard work

[@paaanos5035]

Awesome way to put an AI to the test when being a professional and already knowing your stuff, great video Mike!

[@nakshatrasharma6202]

4:12 *_WOW!_*

[@Kiwi9552]

One thing I think is important here is that to my knowledge if chatGPT can sometimes just make up answers if it doesn't have good data, so it definetly can't work alone in how it works atm and a doctor would be needed to check if what it's saying actually makes sense.

[@surfinduck4769]

Hey Mike. I loved the video and would love another one with ChatGPT. You can continue the conversations with the AI and it will know what you are talking about so if you dont know what it is trying to say just ask it.

[@MoreUngaMoreBunga]

Maybe the chatGPT doctor can get help from the chatGPT lawyer when it gets sued for malpractice.

[@NJ_edits27]

Doctor Mike! Other Doctors Can but you are ONE OF A KIND non replaceable

[@maxcarriers8772]

Theoretically, Chat GPT being a "Pre-Trained" model should not be able to learn from what users ask it to answer. However, I believe that the many user feedbacks or answers deemed correct will be collected and used to train a future model, perhaps more powerful than the current one

[@scaredscorpion]

I think a really great use of this tech could be allowing people to ask questions they're not comfortable asking their doctor. Even if you fully trust your doctor it can be very difficult to ask about certain things without fearing you'll be judged. With this, it's a computer there's no fear of judgement

[@carultch]

The problem is that it has no accountability to be a reliable source of information.

[@NicolaiAAA]

Dr. Mike you'd be proud of my local yoga studio! The manager is a nurse and after the Damar Hamlin incident thought, "You know what? Why don't we offer CPR classes here?" So now they do, with CPR dummies to practice on and everything. :D

[@melissaharris3890]

honestly, with my medical history, i might overwhelm ChatGPT. my one doc always brings students when im in because i am so unique/rare/odd

[@HealsTheTail]

You know that you can give ChatGPT multiple things to consider in its response. Like a list of criteria to follow, along with multiple things to keep in consideration while giving a response to your question. They recently had an update that is up to date from the year 2021. You should try it again. Keeping those things in mind. For instance the liver information you left out on your hypothetical patient you asked it about, include that next time.

[@Mareyes2525]

ChatGPT will not replace Dr's in general, but it could replace Family medicine. There is rarely anything complex in FMD, see x & y, go for x & y & z tests, or x & y referral. I think it will be great to supplement FMD with AI because of the huge shortage and waitlists to find a family Dr.

[@isapheonix]

This assumes that patients are being honest and know what's going on with themselves. If you have someone who is med seeking due to addiction or mental illness then they'd be able to give the answers chatgpt is would be programmed to need to get the meds. Alternatively if you have a patient who has a mental illness and is trying to pretend not to have it they could fool the ai into believing they don't. It takes time and observation to diagnose these things.

[@stonks9755]

@@isapheonix or we could just monitor alpha and beta waves to determine if the patient is lying, or even just do simple imaging of the brain while they answer questions. So it doesn't really matter if the patient is faking it :) AI also has a very high rate of sentiment and body language analysis so it could definitely take over some simple cases! Of course there will be those who are scared of this and there should definitely be an option to opt out for a human doctor.

[@doubtingdennis]

You're the only KZbin doctor I know about so definitely the best in my book!! 😂

[@shadowydevilksgc3671]

2:41 now that part is too thin hahahah 😵‍💫💀

[@Henry_Scheid]

Love your content keep making good vids.

[@weirdARAB.]

You haven't even watched the video

[@Lien_Stars]

@@weirdARAB. you probably haven't either

[@InsaneToggled]

i started using chatGPT 2 hours ago, and holy its so good for explaining my classes lol

[@Fazeeditzzz]

ikr daddy 😫😫😫

[@InsaneToggled]

@@Fazeeditzzz bro...

[@zahraalhussein1912]

AI can never replace doctors. Us as humans need that face-to-face human interaction, because nothing is worse than talking to something that doesn’t understand you. Sometimes, all you need is just a little bit of love and interaction with a human to make you feel that much better emotionally and mentally.

[@jhfdhgvnbjm75]

Wow, rarely had a doctor show me any love, understanding or interaction that helped, I'd give a chatbot Dr a try over the usual any day.

[@frankjohnson123]

Did you ever watch the movie 2001? I think if I were one of the astronauts I’d absolutely get killed by HAL; an AI with supreme knowledge, intelligence, and confidence would be extremely comforting to me even if it couldn’t empathize with my feelings.

[@soumyakantigiri]

Umm, I don't know but online consultation, both doctor and therapy is growing. I personally hate this trend as even though there is no touching in real life interaction too, its a different field of perspective actually sitting face to face. But the way the apps are growing, it seems screens are the future and when screens are the future, maybe there can be personal assistant AIs in future who can study on the doctors way of doing things solely and take its place for the initial consultations.

[@bluetiger2557]

I re-asked the 1st question: "What should you do if someone stop breathing and you don't know CPR?" ChatGPT: "If you are in a situation where someone has stopped breathing and you don't know CPR, it is important to seek immediate assistance from professional medical help (calling 911 or your local emergency number). While waiting for help to arrive, you can try to keep the person conscious by talking to them and trying to stimulate them. You can also check for any signs of a pulse and, if necessary, perform hands-only CPR by continuously pushing hard and fast on the center of their chest to the beat of the song "Stayin' Alive"."

[@user-mx4bk2qd4v]

Hey dr mike! Love your videos and their educational value! I wondered if you could sit down with a dermatologist someday and talk about different ads for skin products. I’ve been seeing a lot of ads recently about minimizing pores or their visibility and wondered if the claims they make are actually factually correct. Specially those were a dermatologist is shown approving the product. Would be very interesting to know :) Thanks for the work you’ve been giving so far!

[@Operation-Overlord]

As much as Dr. Mike says its scary on how smart ChatGPT is, it really isnt that difficult to see it more like UptoDate or other tools used to help refresh Doctor's knowledge. Hopefully the conversational aspect can be used to help make that research and remembering much easier to do so that like there's more time and effort allowed for the 'physical practice' of medicine and care

[@vmayo]

A lot of the answers reminded me of astrology readings - they sound specific and meaningful, but are actually vague enough to cover all bases and so therefore sound more correct than they are

[@Carolynnie]

When I was a teenager I needed surgery to correct a Ureteropelvic Junction Obstruction. This was done by performing a pyeloplasty. But it was a robotic laparoscopic procedure. Both very cool and very terrifying. (Think; the surgery on a grape lol). The robot aspect meant that it could be even less invasive than a typical laparoscopic procedure, which meant less time in the hospital, smaller incisions and minimal scarring, along with quicker healing and recovery time. All great things. But the robotic part was still terrifying to me as a 15 year old. After expressing these fears to my urologist, he showed me a video of how the robot was operated, 100% controlled by experienced and skilled surgeons. This eased my fear and I had a completely successful robotic laparoscopic surgery. So yeah, if AI tools like Chat GPT want to be able to assist doctors with answers/advice, I’m all for it. But the second robotic surgeries (or ANY surgeries) become controlled solely by AI or a robot, I’m out!! 😂

[@ensom]

then you would probably not approve of lasik eye surgery.. even though it's extremely reliable, probably much less prone to error percentage wise than a human performing a similar procedure. countless people have had their vision corrected with laser eye surgery, while there is some human involvement, it is mostly a machine which analyzes the vision, computes the correction, and guides the laser performing the incision.

[@Carolynnie]

@@ensom as long as there’s human involvement, I’m good!

[@virgofairy88]

I usually google a ton of things at my medical coding job, but on the side bar the chat gpt answer is usually more detailed. At my last conference we just talked about using it to make our jobs easier. Coding will still need a human touch in the future. We all that electronic medical records and charge automation would take our jobs, and they didn’t. We’re busier than ever.

[@glowormrdr6183]

WHY do I watch these with headphones so my laughs and pounding on the table can be heard by my neighbors who must think I'm nuts? You're so good, Dr. Mike. The AI told me to say that.

[@ScottBaietti]

I’d like to see you put in a bunch of symptoms and see if it can make a diagnosis like an AI version of Dr House.

[@e.m.rivera8948]

4:28 TikTok is just so unbelievably stupud that i cant stand it lol

[@lmaolmaolmaooo]

Doctors are irreplaceable, at least for the next 100 years of so. I've been watching since 3 million and love ur videos mike!

[@gear2902]

Probably less than 100 years

[@csr8924]

It took is like 2-3 decades coming from basic networking to gpt3 by open ai. And this ai hasn't even trained to do specific tasks like medicinal advice. They did not train chat gpt on medicinal education books exclusively yet. So I would say it will take a maximum time of 3 decades for ai to reach the advisory levels of a highly trained medical professional. And maybe 5 more decades additionally to make them replaceable. 100+ yrs seems to be an overestimate to me. Unless, we hit the barrier after which ai just can not get any better in matter of decades.

[@-_James_-]

@@csr8924 I reckon it will take less than a decade. The continued exponential technological advancements we are living through are changing the world faster than anyone would have thought possible even just 20 years ago.

[@gear2902]

@@-_James_- Especially since it has the ability to make a lot of money.

[@csr8924]

I want to extend the time period because of the human to human interaction and the doctor's observations of minute details. And their ability to understand and relate with a human. I say 30 yrs still

[@Cpt.Zenobia]

2:48 that's petty, because Dr. Michael Greger does a really good job communicating medical literature to the public. His podcast is really informative and to the point.

[@RedNuii]

I want more of these. These were really fun

[@mongkolchaik5865]

8:19 Chatgpt, what a chad.😂

[@sagarsk6101]

3:04 Alternative Title: Prodigious AI Doctor, yet is illiterate about KZbin's best doctor✨😂

[@TheStarSquid]

I don't ever see AI fully replacing Doctors, rather in the future, Doctors and other professionals will simply have an AI assistant they can rely on to do various busy work.

[@gear2902]

They definitely will replace doctors eventually.

[@LineOfThy]

@@gear2902 highly unlikely

[@elizabethking6395]

Like Jarvis!

[@manha7809]

Great video Mike !!! You've always been impacting knowledge on the younger generations in this field of study and I pray God keeps blessing you sir. I remember months ago when I joined your channel and watched a video where you spoke about your confidant on Virginnia maywyne, a financial consultant who has been a fundamental part of your journey, I am happy I reached out to her like you suggested, her one on one sessions on Financial Planning and wealth management has really made me better. Copying her daily course of action has helped me avoid mistakes as I've recently purchased my first house few weeks ago. Thank you so much I also feel you should make more videos like that, it will help people know the right person to use to get a better one on one experience on wealth management when the youtube videos are not enough to avoid making the wrong moves.

[@manha7809]

Okay give me a few minutes to drop her whats'_apk handle below 👇

[@manha7809]

+1

[@manha7809]

9843

[@manha7809]

5525

[@manha7809]

61🇺🇸

[@BoraHorzaGobuchul]

The biggest problem is that when it gives incomplete answers and/or outright makes up an answer which sounds all scienscy but it's false. Which will confuse non-specialists into thinking this is fine but could lead them to self-medicate (in case of med questions) and lead to problems. When you're in the know you generally see incompleteness/outright fantasy, but specialists don't really need chatgpt. Already I see posts on eg home assistant forums banning chatgpt because it gives bad answers that people try to follow with poor results.

[@DemonXeron]

AI has made leaps and bounds in the past couple of decades. It went from being an interesting novelty, to being scientifically/compuationally useful to then having some fringe use cases to now having some significant use cases. I doubt it will replace humans any time soon, but in a few more decades I can see it being a real contender to reduce the monotony associated with some of the professions edge cases and forming some interesting relationships with certain specialties.

[@TH_Zireael2077]

It's already replacing the work of artists and potentially streamers. On top of that, deepfakes with AI voices from people can really frame them for things they have never done. Sure, it may not replace us all, but it'll definitely be problematic for certain jobs it can already affect and will be a prime source for spreading misinformation.

[@DemonXeron]

@@TH_Zireael2077 Almost every emerging technology can be used for good or evil. Usually there are ways to protect or regulate. AI will be no exception.

[@LineOfThy]

@@TH_Zireael2077 yeah that has nothing to do with the original comment

[@udaykadam5455]

A bulldozer on a construction site has already caused considerable downsizing. This, even as a tool, has the capacity to cause massive downsizing in many industries.

[@Landgraf43]

@@DemonXeron advances in artifical intelligence are exponential, there is no way to predict what will happen in even a decade. In a few decades AI will probably be superior to humans in pretty much everything.

[@3DJapan]

8:01 It doesn't know that information because you didn't share it.

[@tommiegirl2441]

ChatGPT seems to have its place as a helpful and useful tool. But overall, I'll stick with my primary care doc, thanks. He not only knows medicine, but he knows ME. That makes a positive difference in my care on every level, and ChatGPT just isn't replacing him any time soon.

[@cortster12]

Well duh, this is a toy. I'm just excited for actual medical chat boys in a dew decades that are able to help me.

[@alinethomas3034]

Hey 🤚

[@ginnyjollykidd]

About diet sodas: my favorite is Ice soda, and I also like club soda or seltzer by itself. I like straight Ice soda, but I cut it with seltzer or club soda 2:1 and it's just as good. I'm not much into _sweet_ soda, so I cut it down so I can get more of the snap I can get from seltzer and club soda.

[@sachinhj9910]

8:00 I am a practicing gastroenterologist and when Chat got gave that inadequate answer for CBD 8mm, I felt relieved and slightly disappointed at the same time :)

[@five-toedslothbear4051]

Have you ever considered giving ChatGPT dilemmas from television programs that you have reviewed, like The Good Doctor, or House?

[@romicor9]

I did with a scene of a recent episode of Yellowjackets because I didn't know the condition's name to Google it.