0:00

[Music] foreign

0:07

[Music]

0:38

foreign [Music]

0:57

[Music]

1:12

foreign [Music]

1:21

[Music]

1:44

foreign

1:52

[Music]

2:15

foreign

2:21

[Music]

2:46

[Music]

2:53

foreign [Music]

3:09

the medical futurist hopefully you can hear me and see me well enough before we start we are in 1996 and

3:18

that's the first time it's the 14th of March and that's the first time I go online I had a chance to browse the

3:24

internet and it blew my mind away I was sure that my life would be different from that point and also that the whole

3:31

world will be different because the internet will open so many doors I had a similar feeling at the end of November

3:38

2022 just a few weeks ago when for the first time I could check out and use chat GPT I had the same feeling that

3:46

it's going to change my life it's going to change the world forever because this is the this is I think the the

3:51

appearance of Chad GPT marks the beginning of the AI era AI is not just

3:57

in research papers and studies anymore AI is a tool we can all use enjoy have

4:03

fears and doubts about but in essence now we all have to start discovering how

4:09

AI can contribute to our jobs to our lives to our fields of Interest that's the topic of today's the medical future

4:16

is live YouTube q a and thank you so much for those being here on the stream so many questions coming in already hey

4:24

guys welcome from all over the world I know that it's quite a global audience

4:29

um and while I love hearing where you are coming from the truth is that I planned for an hour long q a list as

4:35

usual but through the Eventbrite website we've received about 50 something questions really detailed thoughtful

4:42

questions about chai GPT plus I prepared so many things I'm looking at my monitors here so many things for you and

4:50

we will plan to we plan to dive into GPT using it for medical purposes so I plan

4:56

to discover and analyze many things with you about this AI tool today plus you've

5:02

been sending in questions already so maybe oh thank you Balin for confirming that

5:07

you can see me and hear me well enough modern science just asked the perfect question to start the whole stream with

5:14

what is Che GPT chegebt is a large language model it was made available by

5:21

open AI maybe you know this company let me share it with you just in a second

5:27

open AI is an AI research and deployment company and they made this tool

5:33

completely globally available um called chat GPT among others those

5:38

they're also the ones behind Dolly and Dolly to the AI image generators among

5:46

many other things but this is the first tool that is really globally accessible and um

5:52

I plan to show you many things about it but the main thing is that chechi PT is a bunch of algorithms machine learning

5:59

algorithms put together it's a really sophisticated let's say computer program

6:04

that were trained on based on two methods one is supervised learning and

6:09

the other one is reinforcement learning so first it was given many conversations

6:15

and it tried to come up with the right responses and then in the reinforcement

6:20

learning stage of the development of chai GPT it received feedback about how

6:27

well it performed on those conversation based tasks but imagine that happening

6:32

millions and millions of times and and it keeps on becoming even better with

6:37

every single query we put into the chat GPT tool

6:43

um so before diving into all that again I want to go through all the questions that you asked but I also want to answer

6:50

those that came in Live Well just very briefly why why I'm doing this why I'm here with you today

6:57

um I lead the medical futurist Team 15 people and we are trying to help um policy makers governments medical

7:03

professionals patients patient organization leaders

7:09

have a better context about digital health challenges about artificial intelligence we every day I analyze many

7:15

news and we try to describe those to them in a meaningful way we don't over hype Technologies we don't create fear

7:21

we want to give the the middle ground and at the same time at a medical futurist Institute we do the same

7:27

through peer-reviewed research I lead a research team and we publish in high level journals about why digital health

7:33

is a cultural transformation how Advanced Technologies from the metaverse to variable sensors can impact the

7:40

doctor patient relationship and even more importantly what role artificial intelligence can and will play in the

7:47

near future of medicine and Healthcare so I spend a lot of time we spend a lot of time with my peers with my colleagues

7:54

thinking about talking about the potentials of chair GPT and other tools

7:59

because it's not just chat GPT we have a bunch of AI tools out there some of them are just in papers for now and studies

8:05

and we cannot access them from outside some others like this tool and generators are everywhere

8:12

so today I not just I don't just plan to talk about chat GPT and use it in action but I also plan to discuss some image

8:19

generators with you the second question I thought would be the best that I should ask is this it is

8:28

coming from and the question is around has it passed

8:35

the Turing test well you know during this is around whether we can describe there are two doors or two rooms there

8:42

is an EI entity in one room or a computer and a person another room and we provide questions and we receive

8:49

responses and whether someone being outside can determine if the AI is in room one or room two it means the AI

8:56

would pass the touring test if there is no distinction no difference between how they perform on answering those

9:02

questions my my back question to you is does it matter if it has passed the Turing test

9:07

I mean if it has passed during test and if it's a AI entity and almost almost

9:13

artificial general intelligence and so what if it hasn't passed it yet but tens

9:18

of millions of people find it useful and it changes fields of Interest or jobs or lives so what so I think the tooling

9:24

test is just simply not so relevant anymore

9:30

um Jyoti Lama hello everyone I'm a nurse Midwife from Nepal I'm not sure how I

9:35

will help my Midwifery practice I'm here to make myself aware on this new topic and a very interesting idea that's a

9:42

fantastic thing to do I don't think that it will impact directly your Midwife

9:48

free practice but I have no doubts that the patients you come across will keep on will start asking you questions that

9:55

they received from jgbt that they received by using other AI based tools and even if you don't have to use AI in

10:02

your everyday practice but you will have to be able to to help patients deal with those those answers that they receive

10:09

from AI Dr Zee good morning be new as west coast thank you for being here uh happy to be

10:16

here good AI based chats help doctors patients select the optimal surgical approach let's say for a total Heap

10:23

arthroplasty well let's think about because that I don't want to just answer

10:29

the question to you in a way that you get a response I want you to to be able to answer similar questions yourself

10:34

without me being here how can how can someone think about a response for that how do surgeons

10:42

select the optimal surgical approach for a total here arthroplasty they have

10:47

access to medical records they have access to Medical knowledge data medical images through radiology and

10:55

then based on the knowledge they get from textbooks I don't say common knowledge because everything is in textbooks So based on knowledge in

11:02

textbooks and new studies with you know new guidelines they can come up with the optimal

11:07

surgical treatment I think all of these parts are can be digitized all of these

11:14

points of information and data can be digital therefore the chat GPT can have access

11:21

to all of them and from there is just a matter of fine-tuning how we can improve

11:28

the algorithm's efficiency in collecting all those data points and pieces of information and put together an optimal

11:34

surgical treatment plan we do the same with medical students essentially they don't expect them to come up with the

11:39

the perfect answers from day one we spend years teaching them and medical

11:45

students spend years studying and learning these these details until the time they are able to make such

11:51

decisions David H has a question a very good question actually and in the meantime

11:58

please keep the questions coming in the more questions you ask the the lower chance I have to go to the Eventbrite

12:04

website and start answering those 50 plus questions but and you know me I always want to go through every single

12:10

question even if I run out of breath here but that's my plan for today so

12:15

I'm curious about the diagnostic power of CHA GPT compared to what IBM was

12:20

doing with their Watson project the comparison is a bit tricky because IBM Watson IBM sold Watson last year and

12:28

Watson is not even active anymore some new product might come out of it in the near future but

12:35

water doesn't exist for now as it used to be but I understand that you want to compare it to a tool we know that's in

12:42

place that exists and and are being used in action I think jpt uh does fairly

12:49

well on um on these questions I actually as a

12:54

researcher I reviewed a paper like two days ago where researchers put your GPT

13:00

onto a medical exam questionnaire and it performed really well like above

13:07

60 even though they removed some complicated questions questions including numbers

13:14

data medical images but still 60 is like a good medical student average and if

13:21

you put something like this into a chair GPT let me show it to you

13:26

like um well that's that's not not the first thing I did the first thing I did was I downloaded the

13:34

what's it name it's 36 page the this one I want to show you something different

13:40

give me a second here I wanted to show you this

13:45

it's 60 page 64. so the case files for Internal Medicine many medical students have to go through

13:53

this during their studies there are many many case files and Internal Medicine here with the utmost details like you

14:00

get a case and then you are being Guided by these amazing authors how to solve

14:06

that case and I thought that I would just copy edit this part I hope you can see that yes you can see

14:12

that put into check GPT and expect a response and the response was actually uh gastroesophageal reflux disease gerd

14:20

gerd as we call it and the real answer here was I think peptical studies

14:26

they're not the same but they're very similar and it's easy for a medical student to to make that mistake

14:32

themselves but actually we can try doing this so we can just look at another case

14:39

like let me see the next one case five it's a really long really

14:44

complicated case with all these medical records and patient data and

14:51

this one new jet I just eat okay

14:57

it seems it's just too much for this if it all all week jgbt has been down

15:04

for me I think the same will happen now no okay it's up but hopefully now

15:11

so it's not always up if I have I've had many troubles with it technically in the last couple of days I guess everyone is

15:18

using cha GPT now now I will come back to that it lasts for the last case it took about a minute

15:23

or so for it to come up with some answers so I will dive into that after it's done until then let me see what

15:31

kind of questions you keep on asking a b it's a name uh your thoughts on

15:38

implementing authentication verification and authorization

15:43

requirements in order to use the the open AI engine I think it's a good question to us of course it's a fair

15:48

question to us in other fields outside Healthcare medicine I don't see any problem with that but in healthcare

15:55

patients lives are at stake so of course we need to verify the information I just

16:00

don't see how it would happen just imagine that one million Physicians are

16:05

asking chat GPT really complicated questions right now and they are getting 10 million responses who would be able

16:11

to physically verify those results even if we could ask Physicians to refine the results and and run studies based on

16:18

that by the time we finish the study jgpd has a new edition a new iteration so I'm afraid this is why is the the

16:26

mark of the beginning of the AI era because from now on only AI tools will be able to verify as such details or

16:32

information coming out of of chat tools by AI tools so uh

16:38

I expect AI to do that for us and that's why I was astonished when just like two

16:43

three weeks uh no four weeks after J GPT was published I I came across this paper

16:50

uh from Google's research team and a research team at deepmind as well this

16:56

is a very detailed very long study about the large language model they built that

17:01

they called multimed QA and the reason why I'm showing you this not because we want to throw through everything but

17:07

every at the end they they've included real life examples of what kind of questions they they included like I'm

17:14

trying to make it a bit bigger for you these are similar types of case descriptions like the ones I just

17:21

included in chai GPT and that was a multiple choice question and it

17:27

performed extremely well on these questions solving these really complicated case cases it's different

17:33

from GPT because this model was trained on medical data on medical databases on

17:39

case studies therefore it's a bit more prone more maybe trained towards

17:45

answering specifically medical questions in the meantime

17:51

[Music] yeah I have to show you the chat gpt's

17:58

response because it arrived in the meantime based on this after like thinking for one and a half minutes I

18:04

guess not because it's that slow but because of the high load that the

18:09

website is experiencing right now it says it suggests the results and the symptoms suggest that she may have hyper

18:16

night hyponatremia which is characterized by a low serum sodium level

18:21

here's the case hyponatremia uh yeah so I think it's a

18:28

pretty good response I mean look I always try to be as objective as

18:34

possible as a researcher about the advance the progress of these Advanced Technologies but

18:40

if it doesn't blow your mind away then nothing will it's it's uh it's far from artificial general intelligence it's

18:46

very much artificial NATO intelligence but man it's it's a breakthrough when

18:52

you provide such a case and it gives such responses um

18:58

it will if your shape how we access the information as positions how patients

19:04

obtain information online I mean two years ago even today there is a

19:09

general discussion among Physicians that what if my patient goes online and looks for their symptoms and they will find

19:15

all these information on pieces of information on medical websites and what we'll do what will they do with those

19:22

pieces of information that that discussion needs to change now what if my patient uses artificial

19:29

intelligence to respond to a very specific very detailed medical case

19:36

and how do you respond to that in the case of websites at least we were able to say that well if the website had

19:43

medical editors or medical professionals as editors at least we know that Physicians are working on the content if

19:49

it's updated regularly if it has a terms and condition and all these things then at least we can say that the information

19:55

in general can be trusted enough to be used in the medical decision making process but what do you say now to that

20:02

patient asking about well AI told me to do this or that what's your response to

20:08

this it's just simply mind-blowing okay oh many questions keep coming in

20:14

please keep them coming in I will go through all of them even if you have to see through the night here at least it's

20:20

uh 4 20 uh in Budapest in Central Europe right now

20:25

all right now I'm trying to find the next question from someone who hasn't asked questions

20:32

yet uh modern science could I be more specific

20:37

about what about chechi PT or some other thing I said rohini Verma how AI work in medicine

20:44

Discovery and result given by Ai and how much it is accurate it was a very

20:49

strange question it was this way but I get the implication of your question

20:54

there's no way to tell how accurate it is like um if you ask a general chatbot

21:00

a question like there are many I mean we we publish this paper this article on medicalfutures.com not so long ago we

21:07

published this every single year the top 12 health chatbots and and you can find so many chat Bots here from other health

21:15

Babylon Health really really good ones we review them every single year and and choose the 12 ones I like the most and

21:22

if you ask questions if you use these chatbots to ask medical questions in

21:27

most cases the responses you get also have sort of reference or further reading

21:33

section so like we made these decisions based on these are these are those

21:39

references medical websites medical studies but in the case of jgpt you know

21:44

what is the reference it's just an answer it's a good one by the way it's a very complicated one but

21:50

it is just an answer and there is no way for now to verify these results and I

21:55

just don't see um a systematic or technological way for verifying these responses I think what's

22:02

going to happen now is that you and I if you're a patient if you're a physician a policy maker if you turn to an AI based

22:09

tool and use it for such conversations you need to be the one verifying the

22:14

information if you use the information it's your responsibility you cannot tell anyone that it was the AI giving me that

22:21

AI tools are amazing they keep on providing such answers but it's gonna be our individual responsibility about

22:28

verifying those pieces of information the responses the details the answers they get from these conversational

22:35

chatbots I know it doesn't sound I mean optimistic or well enough

22:40

but I see the the Romantic value in it that we have to

22:46

get better individually to make a distinction between information we get from AI tools which

22:52

are good which are bad or or information we get from analog websites not even AI

22:57

tools Andre Olivera it's a good question to ask actually what is the best used to

23:05

start a few days ago I had my knees over she's 12 she's 11 years old and I made

23:11

cocktails for for the guests we had at home and and I wanted to make a non-acolic cocktail of course for her I

23:17

had no idea how to do that so I just asked chiji pity well can you please make her I always say please when I told

23:23

you Technologies can you please make a non-alcoholic cocktail recipe for me using these ingredients and I listed

23:29

what I had at home then it came up with a brilliant idea so I don't think we should use it as a start for on

23:36

answering medical questions and and solving medical cases ask something you have been thinking about the other day I

23:44

let me um oh you can see it now let me go to

23:49

the browser you can see my brief okay you can see my

23:56

you think I give up that easy here is my brief so

24:01

um can you almost there can you please

24:08

um come up with ideas about

24:13

what kind of new videos I should

24:19

make for my YouTube Channel I know you don't like the

24:24

capital O uh that is about the future of medicine digital

24:31

Health Technologies and uh use of artificial

24:37

intelligence in Healthcare

24:43

I'm always very very kind that's what you say I do so when there is a you know Skynet like scenario I hope that they

24:50

will be very kind to me so in a few seconds hopefully we would see the response coming from chat GPT

24:58

and why we are looking at that I cannot check your new questions but please keep them coming in

25:04

um so yeah start with something easy something about your life something you've been thinking about I the other

25:11

day I asked chpt to come up with a birthday gift ideas for my wife who is that many years old likes this and those

25:18

things and Robbie has a daughter and so on and and it came up with really good responses really good ideas I I didn't

25:25

expect to see okay we don't have that much time check GPT so I will come back to this in a few

25:32

seconds like in the other case but in the meantime I want to go on with your amazing questions

25:38

all right area Muller if the question is for me

25:44

what are you doing now professionally I'm a researcher focusing on digital health and Artificial Intelligence being

25:49

used in medicine and Healthcare you can find my researcher profile on

25:54

Google Scholar obviously how can we be reassured that the data

26:02

used for chegebt is scientifically valid you cannot be sure about that Chad gbt

26:08

doesn't even provide you with the references based on which it made those decisions but open AI think will provide

26:16

more details about how chat GPT was trained but now as I mentioned before we only know that it was trained on

26:22

conversations through two methods one was about supervised learning and the other one was about reinforcement

26:29

learning but just a general Essa

26:35

a general it's it's never in the middle I'm the

26:42

only one being disturbed by that okay I'm doing my best for you now uh just a general reminder that based on

26:49

Nick bostrom's amazing book super intelligence there are three levels of AI if you zoom out and look at the

26:55

long-term perspective of this technology what we have today is artificial narrow intelligence meaning one algorithm can

27:02

do can perform a very well defined single task but one single test with an IQ of zero the next level could be

27:09

artificial general intelligence when one algorithm has a human being's cognitive

27:15

capacity so in short whatever you are capable of that AI I say entity not just

27:21

the algorithm would be able to do the same and the plan let's bring AI

27:27

researchers even those working at open AI have in mind is just not to reach artificial general intelligence and now

27:34

we are still in the artificial narrow intelligence part I would say somewhere here maybe we were here three months ago

27:41

and we are here now but we are really far away from

27:47

artificial general intelligence no matter how impressive um chai GPT is

27:52

right now couldn't come up with I want I want 10gp

27:59

to answer the question about my YouTube channel but because the oh it's coming it's coming it's happening let me show

28:04

it to you so ideas on my YouTube channel series on

28:09

the latest advancements in digital Health Technologies in telemedicine where I was good good idea I had one like this before a review of current and

28:16

upcoming AI power diagnostic tools I love that I haven't done it yet I've been I've been producing many videos

28:23

about AI in medicine but not you know about especially diagnostic tools a

28:29

discussion of the ethical implications of AI an overview of the role of blockchain technology in healthcare a

28:35

look at the future of personalized medicine like genomics proteomics a review of current state of subject

28:41

current state of Healthcare in developing countries wow it's a great great topic the exploration of the

28:47

impact of virtual reality and other types of realities AI in drug Discovery and simply

28:53

beautiful I mean I could come up with such things because I've been producing like 500 videos on my YouTube channel

28:59

for 10 years but to do it in like 10 seconds is just simply extraordinary

29:05

in the meantime I come back to your questions which keep coming in

29:12

so you can't be sure that the the responses you get on share GPT are accurate you have to validate them them

29:19

yourself yes you make a good point that jgbt the

29:24

next iteration will have a professional Edition and if you pay for that you will have access to it all the time plus if

29:31

you think about it then uh just it's just a thought

29:37

how much jgp can do in practice right now that's running in the background you

29:43

know in the research Labs of open AI because what we have is a tool in the

29:48

open public but in the background they must have an even more sophisticated

29:54

more Dynamic even smarter addition what can that addition be capable of that

30:00

just simply mind-blowing okay how to teach chpt is asked by Andre

30:07

Oliveira I think it's being constantly done not just by open AI using

30:12

reinforcement learning methods but even if you come here I go back to the chat GPT answer

30:18

here it is you provide feedback and I think these kinds of feedback matter and

30:24

you keep if you do that you keep on contributing to uh chair GP these Advanced I don't think that these

30:31

written feedback forms will be the ones uh really contribute to the major progress of jgbt but I think it these

30:38

are being used heavily by the researchers working on that

30:43

uh do you think Google deepmind will better than chat GPT for specific questions I do think they have a chance

30:50

to be better than charge GPT so for medicine and Healthcare multimed QA can be better than jgpt apparently it is

30:57

better based on the study I just share with you actually I thought we could run an experiment here if I go to the study

31:04

that Google published um you know these are the these are

31:09

these These are these questions are in the supplementary materials and they

31:14

provide a case and then they provide for potential answers and I

31:22

let's be very bold here and I will include that question for chat GPT

31:28

without providing a chance what which what I will I will just ask what is the

31:35

next back step in the management of this patient and we will see what it comes up we don't know yet

31:40

and I will try to compare the results browser

31:46

it will be the next best step in the management of this patient would be the consultative psychiatrist and a

31:52

perinatologist a specialist in high risk pregnancies to determine the best course of treatment is to involve adjusting

31:58

your current medication regimen close monitoring yeah yeah to establish a regular contact

32:05

okay so it's very vague because there are no specific treatment options being

32:11

offered but actually it just made me curious whether what if

32:16

I provide some potential responses like what they did with the Google tool

32:23

the multimeter QA will it be able to find that one

32:30

no come on to GPT we have been kind to us throughout the whole stream let's keep on doing that

32:42

ah okay we will come back to this I will monitor it for you but in the meantime I

32:48

want to keep on going with the questions uh Dr Klaus Christopher amalunk do you

32:54

pay for Chad GPT or do you get this for free every difference to Google and choose the

33:01

proper answer well I don't nobody is paying for that now but there will be a professional addition which means you will be

33:08

guaranteed an uptime when you want to use chat GPT you will have access to it no matter how many people are trying to

33:14

use it it's an understandable business model and the difference between Google and jgbt is that Google gives you the

33:21

best um guess the algorithm has about which

33:27

website or link might have the response to your question or the information

33:32

you're looking for opposite to that chagibit is trying to answer that question as a conversational

33:39

agent it still get those information throughout the whole internet from before 2021 so nothing new nothing

33:47

refreshed maybe do decrease the chance of it being used for the wrong reasons but it's a conversational

33:54

uh chatbot oh I'm smiling because it just started

34:00

responding to that it it has been thinking for about more than one minute and the response it gave is B

34:09

so it understands it understood that it had four choices and it made the choice

34:15

and then it describes the reasons behind that in the Google paper the the answer

34:22

was D actually and there is no reason provided for that

34:28

so I would say it's a it's a bad choice you see how can you make how can you

34:34

make a difference like uh you come up with such a complicated medical case and you start inputting them into medical

34:40

chat Bots and those will come up with responses but even if you verify that the Google chatbot was giving a great

34:48

response but the chat GPT was not useful it was one question but there can be

34:54

billions of questions and how can you verify each of them every single time you want we will be the ones modifying

34:59

each of these questions no matter how complicated Future IT entails

35:06

oh doctor science of course it's a chat actually let me just include the link to

35:13

the actual chair GPT here this is the link all right

35:19

I go back to the last questions thank you Jay Polizzi you need to create

35:26

an account at open Ai and then you have to log in and there you can use the chat

35:32

Gustavo zumelzo how do you think it can be used by help personal to reduce

35:39

administrative tasks overload I'm so happy that you are this question because we actually published a a article about

35:46

it like one week after church GPT was really coming out and becoming a

35:51

widespread and available and we came up with a few like scenarios such as it

35:57

would summarize medical records based on patients family history symptoms and lab results just imagine that the patient

36:03

comes in there is one paper they bring in with the medical records they have one paper

36:09

about family history and one paper about symptoms and what more about lab results if you put all these together you as a

36:15

physician have to be able to sit down and write everything down into a condensed medical record profile which

36:23

you could do that for us and you only have to reconfirm that the right details were written down also I hope that you

36:30

can see yes you can see you it could summarize analyze research papers like uh chagibility please list keywords for

36:36

me in an abstract or summarize a long and detailed research paper before you ask let yes it means that I think it's

36:42

going to be possible to write full research papers using chat GPT

36:49

um yeah let's I think a whole different live q a session should be dedicated to

36:54

that alone we think that jgpt could be could be useful for writing General texts like

37:01

emails or a blue blurb like basic descriptions like there is a

37:06

the flu season is coming up and you as a physician you like sending out the generalized emails to your patients

37:13

about raising awareness about the importance of the flu vaccine kgpt can do that for you in three seconds it could answer broad questions

37:20

like you are looking for something but you're not so good at Googling you can

37:25

ask questions about that like what is the I don't know health insurance code for this or that disease in this and

37:31

that zip code and it will be able to come up with a response you could also work as a like an administrative

37:38

assistant like it we also ask it could you work as a chatbot in a doctor's office and

37:44

describe what kind of things it could do I think the last thing is that the thing that you all should start with it it

37:50

helps you build a relationship with AI it makes sense believe me you have to have a relationship that you know that

37:57

you are aware of that you know how you respond to such queries how you respond when you see the responses coming from

38:03

an AI tool because it is it will Mark and determine the way you will keep on working with these AI based Technologies

38:13

um Ricardo bezo was the very good question is check GPT using all the information

38:20

presenting the web at the time of the question no just before 2021 to avoid

38:25

focusing on recent events meaning the question answered today by GPT could be different in a few weeks

38:31

after the information on web updated no because everything is crawled from the web before 2021.

38:41

sensor for how it can be used in healthcare it's a good question let's ask our

38:47

beautiful friend called chat GPT here it is you again again you can see

38:55

that but you will in a second here it is let me ask the question um

39:03

okay oh how Chad GPT how chat GPT could be used

39:11

in healthcare I'm very curious chat GPT can be used in healthcare in a

39:17

variety of ways such as virtual Health assistance charge EPT could be used to build virtual Health assistance okay

39:24

medical documentation it could generate medical documentation such as discharge

39:29

summaries and progress notes amazing critical decision support it could help provide clinical decision support by

39:36

analyzing patient data and suggesting appropriate treatment options medical education to create interactive and

39:42

personalized learning experiences medical students telemedicine it can be used to facilitate virtual consultations

39:48

through natural language processing medical research it could assist as I mentioned researchers in summarizing

39:54

large amounts of literature and critical trial matching wow it could be used to

40:00

match patients with appropriate clinical trials based on their medical history and symptoms

40:05

guys I would love to be able to say to you that these are just fancy responses

40:11

and it's just too good to be true but all of these are could be used in and

40:16

could be could be working in practice it's simply mind-blowing uh Solomon is it possible an IOS app

40:23

version I don't think they have an app for now it's just a website where you can get involved with the conversational

40:29

chatbot would also be great to have a real-time transcription of like a 50-minute

40:35

therapy sessions and have chatgpt create a quick summary let's take fire by law

40:40

if you think about that um a future iteration of charge GPT I

40:45

think could be could be able to do that is it considered to be specialization according to Medical departments some

40:52

and I don't understand the question please um elaborate on that

40:58

it has the same answer like hierarchical I guess it keeps on providing the same answers uh for the same questions why

41:05

would it change because all the information it gathers are before uh 2021 I also want to share the article we

41:14

wrote about it it is here and let's keep on going with the questions

41:20

um either adjuster quite a regular

41:25

guest on this channel had this question what might be the next few years evolution of AI enabled medical chat

41:32

recognitions to take on to move Healthcare forward for example if AI enabled medical chat became increasingly

41:37

integrated into a health sensor ecosystem I just think that because of Dr

41:44

shortages because of the 5 million Physicians missing now in healthcare globally and the 10 million that will be

41:50

missing by 2030 at some point maybe we are very close to that point every stakeholder in

41:56

healthcare will have to accept that we have to reach out to GPT or similar AI tools to fill in the gaps it's better to

42:03

provide information that must be verified thus we cannot know for sure if

42:09

it's good or not before verification compared to providing nothing because that's what we do with at least a

42:16

billion a billion patients today we are not able to provide any kind of care

42:21

because we simply don't have enough trained medical professionals to do that so so chatbots in that way will jump

42:27

into Healthcare and become implemented I think even from this year I guarantee you that there will be Healthcare

42:33

Systems using charge GPT in action uh

42:39

Orleans had this very smart question I think the last part is the one I haven't

42:44

answered before what possible ethical problems could come from the use of charge GPT I can only give you ethical

42:51

problems for now uh it provides a wrong response to a physician who makes a bad decision based on that because he or she

42:57

wasn't able to verify the the quality of the information it provides a bad decision about

43:03

magnification which leads to side effects or hospitalization it gives patients bad recommendations or

43:09

lifestyle insights it summarizes a research paper in the wrong way it summarizes the medical records in a

43:16

misleading way which would be very very hard for a human physician to spot like

43:21

it makes an error that of course not intentionally but it's an important detail that's different from the real

43:28

data patient had and the data that got into medical records that might lead to bad decisions later on I can go on

43:35

forever I can only think of ethical issues and considerations

43:40

for now Eric Van Der Holst has a very smart

43:45

question to ask how will bias be corrected in chai GPT for example millions of old school treatments

43:50

compared to new insights that have little presents on the internet I think no way even the best AI dues can be biased

43:57

because the database behind them the databases they were trained on were biased and it's only it can only be the

44:03

conscious decision of those training AI based tools to avoid bias by making sure

44:10

that the databases are unbiased I don't see yet how AI would be able to correct

44:15

those buyers um

44:20

how you compare from the medical point of view GPT and MD Palm like what is MD

44:26

pump let me check it's a chatbot If You by MD pump you mean the chatbot

44:34

Google came up with then I would rate it much higher uh on comparing it to jgpt

44:40

on um responding to specifically medical questions

44:46

uh modern science can gen GPT be used in nanotechnological devices that focus on

44:52

Healthcare and none of pharmaceutical drugs I guess for asking jgpd questions

44:58

about the research components of that it could be useful but I don't see how it would change the field for now

45:09

it's a good question to us that's why I'm smiling how can we make people desist from asking to chat about instead

45:17

of doctors should a hard limitation be implemented like the banner that pops up with copied information

45:22

I guess there is something like that when you go to the new chat look at this they have amazing recommendations here

45:30

Nate occasionally generate incorrect information May occasionally produce harmful instructions or bias content

45:38

limited knowledge of word and events actor 2021 what else a pop-up would help

45:43

this is just I just we all have to acknowledge that this is using these AI

45:50

tools is not fundamentally different ethically from using a website to answer

45:56

a medical question or to look for medical information it used to be the the final decision of Physicians and

46:03

patients about the medical decision making process even if when they used to just check medical websites now they

46:09

will use AI tools and will use recommendations coming from AI Based Services that check all the data coming

46:15

from all our health variables and smart watches I see the the difficulty in that yet

46:21

it's gonna be people at the end of the line making decisions they have to be the one verifying all these details I

46:28

know it sounds like a this topic scenario but I always look at this from the other side that

46:34

we get a chance to um really summarize the information we needed uh from the internet that we had

46:42

to build skills for years and years to get better at in a few seconds without

46:47

having those skills that that that's the only thing I see with the rise of these chatbots

46:53

all right next question uh

46:59

it's a very long question I don't even I'm not even sure you can see the whole okay okay you can see the whole as only

47:05

as simple Yes before 2021 uh knowing how greatly medical information is doubling thoughts on how

47:12

we get to a point where AI can better understand the medical literature and know what studies

47:20

and use the information to help Drive clinical decision making or Google has a

47:25

healthcare chat GPT in the works yes um what should be the goal here to build

47:32

chat Bots that can answer all the questions Humanity might have in mind or to provide a support tool that every

47:39

physician has been dreaming about I think that these AI tools can fall into the latter category and I don't see

47:47

why we will focus on Building Services and chat Bots that could answer everything because then what's the point

47:54

in having us what's the point in going through six years of medical training and then who knows how many years tens

48:01

of years decades of specialized medical training I think this there's a support

48:06

that the tool should be a support tool that could contribute to our medical decision making not something that does

48:13

the task instead of us moreover it's not able to do that and not even in the

48:18

coming years would it be able to do that all right will it help public health

48:23

professionals Us by Tariq that's the plan yes as I mentioned before

48:28

um that's a good question what new digital Health Ventures can be created with chat

48:35

GPT the questions coming from Andres Cardona Mendoza

48:40

well um anything that helps that helps summarize information uh research

48:48

website that summarize abstracts and really complicated studies and clinical trials and research papers

48:54

services that would summarize medical records examples I described before and

49:00

you can build Ventures on that but for what reason when chain GPT can do the same for free in seconds uh a few days

49:07

ago I was at a dinner and the CEO of a company told me that he has a binary and

49:12

he told me don't tell me that this AI tool could design a Vine label from my

49:18

upcoming Rose Vine and I went to me Journey on Discord and I designed a line label for it actually if you want to

49:26

have some fun here we can try to do the same we can go on

49:32

chat GPT to the Newbie section and you again okay now you can see my brief

49:38

slash imagine and my prompt is uh violable about a

49:46

non-alcoholic wine with very

49:51

colorful and I don't know modern colors Arena photo realistic

50:00

style I don't know I just thought that people should give it a try and we will

50:05

see what meat journey and AI image generator comes up with uh

50:12

uh you think I can see my my label yet it

50:17

lost oh it's coming it's coming it takes a few more seconds

50:26

maybe a few more 62 percent it's gonna be 90 something and then ready or

50:32

non-alcoholic Vine label with very colorful labels will be ready in about

50:38

two three seconds it will get vanished from here it will go to the bottom of the page

50:43

in a couple of seconds that's it it's ready

50:49

here it is

50:54

it it provided even a name a very silly name Will okay but it's it's beautiful design it's

51:02

spectacular so yeah it's possible to do even such

51:07

things with AI based tools actually if you like if we were um uh

51:16

you know what let's keep on playing with me Journey for a few

51:21

seconds I just launched a new medical practice and design uh not just a

51:29

logo for uh Medical Practice that uses

51:37

Advanced Technologies likes um

51:43

bright colors and uh is very

51:48

Petty futuristic well we will see but using AI tools I actually I we have

51:54

been working at the medical features an article about that that using AI tools a

52:00

physician today could create like a logo with an AI tool that is my oh here it is

52:07

an AI website with AI tools content for the website to copy editing tools copy

52:12

edit for the website images uh video content even through deep fix a lot of

52:18

things just by using free AI tools I even have a tool called voice.ai with

52:23

which I can change my voice and I sound like Morgan Freeman you can do all these things now with

52:28

free tools it's really mind-blowing and it's it's getting harder and harder to keep

52:34

up with the pace of change about these AI tools it's my job as a researcher to

52:39

do that and it still takes me hours every single day now come on the logo is almost ready man it's gonna be a great

52:46

logo it's in the bottom somewhere where is it here it is

52:52

Jesus Christ well this brain looks amazing the the text is not good enough

52:59

because I think median is really bad at putting text on images but this logo or

53:04

that man these look spectacular

53:12

if you want to do something in in a no let me go back first if you want me to

53:19

design like an image for uh for a healthcare based practice then just let me know and I would try to

53:26

do that on me Journey

53:32

Dr Z why is chat GPT up to date only to 2021 because it makes sense to avoid it

53:39

um reflecting on recent issues recent events it could uh it couldn't make a

53:45

good decision and just restricting it to information before 2021 it it doesn't

53:51

allow people to make predictions based on that or you know uh bet on sport

53:56

teams based on the the responses they get from chat GPT so it's a bit harder to use it for the bad uh or wrong

54:03

purposes wrong reasons Andre elevator can I say that Chad GPT is more wealth structure chatbot

54:10

conversation I guess you mean compared to other chatbots on the market

54:17

no I think uh other chatbots on the market uh work like or like an end and

54:24

this one is a lion it's a bit more complicated it can do so many things it's much more dangerous

54:30

I'm sorry for the bad metaphor but you get a point

54:35

uh my best let's yeah you can ask your GPT if he thinks AI will be replaced by

54:41

nanotechnology actually you just gave me a good idea here let's ask

54:47

veal artificial no let's ask the question everybody is

54:54

always asking me real artificial intelligence replays positions

55:01

right AI has the potential to assist physicians in a variety of tasks

55:09

Diagnostics treatment planning that is unlikely to fully replace thank you Chad

55:14

GPT for confirming my point that I make almost every single week

55:19

my AI can process large amounts of data and make predictions based on the data it doesn't have the same level of

55:25

understanding and expertise man it gives a better response to that question about itself than a human being would be able

55:32

to that's uh that's spectacular

55:39

Adam Schmidt thank you and welcome to the to the stream will this make patients who use jgpd for getting

55:46

information about their condition question the MD's knowledge absolutely but that's the same again with patients

55:52

going on websites and finding information that they bring to the next Doctor patient visit they will question

55:57

their knowledge they will ask the specific questions it has been going on for two decades now ai will make it a

56:04

bit more complicated but I don't see that fundamentally how these two things are different now

56:09

uh Sandra schluchter how do you avoid influencing of the data you cannot

56:17

it closed the entire worldwide web so how can you make sure that the data you change will be the only data point jgpt

56:24

looks at I don't see how it could be done

56:30

oh a good good question here about copyright

56:36

who is the copyright for your Vine label anything that you create through these

56:41

AI tools nobody has the copyright for them you are free to use as well as anybody else is free to use unless you

56:48

pay an amount and then you are the only one who can use it

56:55

uh Dash logo it's a good logo come on it

57:01

was beautiful oh how to feel react when we discover a

57:08

new disease so under the Oliveira it won't understand the question um

57:14

he doesn't know how we feel doesn't know what it means to discover a new disease not even a physician will be able to

57:20

answer the question I think

57:25

foreign and you want to see an elegant and

57:31

minimalistic Psychotherapy practice logo including a lioness Jesus okay man I

57:37

said I would do that for you so I will do it for you if I promise something I deliver guys

57:44

imagine uh an elegant and minimalistic

57:51

Psychotherapy Medical Care it needs medical is a important word here practice logo

57:59

that includes lioness okay we'll see it's a very strange brief

58:08

and you are more than welcome to come up with similar briefs I have many more questions to go through us so I think I

58:15

will not stop at the the one hour mark we will keep on going into the second hour there are so many more questions to

58:22

answer we are getting there

58:27

it's going to be a logo it is a lioness so uh it's not totally unbelievable that

58:34

it will be able to do that it's done that's it

58:45

well I would maybe this could be a psychotherapy but the medical part doesn't come through to me it's just a

58:51

logo about a lioness so it was not it was not as impressive as the ones before

58:58

this uh the site J police that I'm using is called mid Journey it's uh let me

59:04

include that in the chat with me Journey it's a Discord

59:12

Discord channel so you have to be a member of this chord it's a free site you have to register and log in and

59:18

after that you can use mid Journey the channel of me Journey for free there are I'm going back here hopefully you can

59:26

see that again but you can see that but I will make you see it there are newbies rooms these are

59:34

the ones that you are looking for and in newbies rooms you can ask these briefs

59:40

all right let's keep on doing with the questions there are so many coming in I'm really I really appreciate your time

59:47

and everything for being here and asking all these amazing questions um because GPT will be considered by the

59:55

FDA or in Europe governing entities as a medical device no way no way not for the

1:00:00

next year or so the FD and the European Union have even hard time approving

1:00:09

medic AI based medical technologies that have clinical trials and peer reviewed

1:00:14

support and and um um evidence in the background so um jgbt

1:00:19

is far from that the first papers are coming out in those in these very weeks about chai GPT and that's just the first

1:00:25

iteration we will see how it evolves in the future

1:00:31

all right

1:00:38

next question what is what if chat GPT is used by

1:00:43

patients and takes and take treatment themselves after getting diagnosis and over-the-counter medication the same as

1:00:50

it happens when it happens through a patient coming across the same details in an automatic website it's the

1:00:57

patient's responsibility that they shouldn't use anything that that visit that and not Physicians are recommending

1:01:03

to them it's just general knowledge if you if you read about a treatment on a billboard out there or on the website or

1:01:11

you get the same from an AI tool there is no League from the legal ethical social perspective there is no

1:01:17

fundamental difference between these three examples all right

1:01:23

what can you say about the recent announcement of the paid version of chatgpt

1:01:29

that I will pay for that I will pay for that to be able to use it every single time I want to use it unfortunately and

1:01:36

that's not just the first AI tool I will be paying for and I think you will also be paying for many others

1:01:44

um use of I'm going through many questions

1:01:49

but we've covered the most cyber threat using medical cases medication management how to use jgpt for a Pharma

1:01:57

company I don't see yet a Pharma company should invest into AI being used in drug design and Drug repurposing not judgment

1:02:03

but you will see jgbt in Microsoft Office tools I think before you see them

1:02:09

being used by Pharma companies uh

1:02:16

is there any tool available free of charge for generating large synthetic biobank

1:02:22

databases I don't think so there are guns generative adversarial networks

1:02:28

that are that are able to do that but not through free tools and and tools like jgpt

1:02:34

I would like to know how to be a power user of chegept and AI not Discovery I

1:02:40

don't see for how for now how you can ask your GPT to come up with information that you could ask also in a Google

1:02:46

search otherwise I don't see but if you want to know let's just ask our dear

1:02:52

friend here chat GPT

1:02:58

how can chat GPT be used for improving drug design and

1:03:07

Discovery for pharmaceutical companies

1:03:14

let's see nothing yet again it's a bit slower than

1:03:20

usual you know it's a free version so I will oh jtbd is a large language model can be

1:03:26

used to assist in raw design okay text Mining and natural language processing

1:03:31

it can be trained to process analyze large amounts of unstructured data texts

1:03:37

scientific articles papers drug pathway okay predictive modeling it can be used to build predictive models that can

1:03:43

identify potential drop candidates based on their chemical properties or other characteristics

1:03:49

virtual screening to simulate the The Binding of small molecules that Target

1:03:54

protein Jesus Christ generative models to generate new chemical compounds and Drug candidates

1:04:01

just some examples with simply mind-blowing it that it can do that so if I if I

1:04:07

included some data from an Excel spreadsheet it would be able to do these things for us

1:04:13

okay next question uh do you say chargpt in society health

1:04:20

education well if it can answer questions about General general knowledge about health and

1:04:28

disease management then of course it can contribute to that but for that it has to be verified which won't take place in

1:04:34

the near future what does it add up to the field of

1:04:40

service design I have no idea uh Desmond Thomas yes we described that

1:04:47

discussed that before that unfortunately we have to be the ones verifying the information coming out of jgpt because

1:04:53

there is no details about how there are no details about how J GPT came up with the response

1:05:00

Philip do you have any discounts for your digital Health course yes we provide discounts for medical students

1:05:06

up to like 70 just send an email to uh to my email address about it

1:05:13

which is virtue.com medical freechoice.com without proof about you being a medical student

1:05:22

um what do you think about the response of medical fraternity to judge beat and other AI tools since these tools are

1:05:29

reducing dependency on doctors I'm not so sure about that and making them less and less important for patients they

1:05:36

just I don't think that's true I think these AI tools being as amazing as they

1:05:41

are it will make patients depend even more on Physicians because as I mentioned before the details they get

1:05:47

from these tools cannot be verified so they will increasingly need their Physicians to help them deal with such

1:05:53

information coming from these tools uh is there a way for chair GPT to

1:06:01

detect typos in the text input I'm pretty sure about that just ask to look for typos in this or that text and

1:06:07

copyright the text into that I'm pretty sure it can do that it's quite a simple test to do even Microsoft Word can do

1:06:13

that not an AI based too all right how to use jgpd as a symptom checker

1:06:20

very cautiously because the same bsv I demonstrated before it can come up with

1:06:26

really wrong answers uh

1:06:31

how reliable is it okay we discussed that before how to integrate in the clinical workflow it very much depends

1:06:37

on at what stage you want to integrate it into the medical workflow like in for

1:06:42

administration or just like sending emails that's something you can do in the background but when it has to

1:06:47

interact with patients that's a different scenario so it's very much depends on details how cha GPT could

1:06:53

leverage Radiology I guess is the other way around how Radiology could Leverage The Power of chai GPT

1:07:02

um actually it's a good question to ask let's ask our friend uh how can

1:07:11

how can chat GPT used in Radiology

1:07:16

let's see at some point I believe in stop answering your questions I will just

1:07:22

include every single question into chai GPT and I will just um correct the

1:07:27

responses it provides if those are not correct then I will correct them but that's all so jgpd for radiology image

1:07:34

analysis it can be trying to process analyze medical images like x-rays memorize okay report generate Radiology

1:07:41

reports based on images it can be used to process and exact information from ideology reports and it

1:07:48

can be used to predict it's very similar to the other question that we asked so it seems that it keeps on providing

1:07:54

similar responses I would love to be able to test it with actual medical data that would I really mean something

1:08:02

uh uh okay next question wow at least 20

1:08:08

more are here how can we use Ai and chat GPT for telehealthcare

1:08:14

I don't think we can use jgbt for that now but we can use AI for creating deep

1:08:19

fake videos or Physicians and a chatbot could answer the questions maybe use chatbot and a deep fake video to make it

1:08:26

look like the patient is having a telemedical consultation I know it doesn't sound ethically fair but because

1:08:32

of doctor shortages at some point we will come to this and I hope you know because we discussed it

1:08:38

I think two live q and A's before what is the trick for you to make sure that

1:08:44

the video The Physician on the video you are talking to is not a deep fake video

1:08:49

do you know the trick for that chat just let me know what do you have to do to find out if your deep fake physician

1:08:57

is a deep fake physician indeed and not a real one in the meantime I keep on going on how

1:09:04

can students benefit from chai GPT this is really a question if I was a medical

1:09:10

student today boy I would be your jgbt like three hours every single day

1:09:17

you know we have the uh the digital Health course and on the digital Health

1:09:22

course uh students everyone who's taking the course I just share the link with you now has to submit an assignment and

1:09:29

the assignment can be a press release about the medical technology going berserk in 2032 or a scientific short

1:09:37

story anything like that and I received about over 1 000 assignments since the end of December I'm not sure if I'm

1:09:44

reading assignments written by those people who are taking our course or assignments that they just ask chair GPT

1:09:51

to write for them in the meantime nobody answered the question about the the Deep

1:09:56

fake trick you have to ask the fake physician to look at their side that's

1:10:01

something deep fake cannot do for now but do you know what let's ask again our good friend here

1:10:08

uh write a press release about the medical technology going

1:10:17

berserk in 2032. that's one of the assignments uh in my course I hope it

1:10:24

doesn't come up with anything useful because if it does then I'm doomed I just won't be able to tell whether

1:10:31

students are submitting handwritten assignments or or not I hope it doesn't

1:10:37

come up with anything useful in the mean

1:10:43

in a shocking turn of events a medical technology designed to assist doctors and nurses in their duties malfunction

1:10:49

today causing widespread chaos and injury at a hospital name according to initial reports a technology highly

1:10:56

Advanced artificial intelligence system was being used in the hospital's emergency department when it's suddenly malfunctioned the system which had been

1:11:03

designed to assist with patient try agent diagnosis began making erratic and dangerous decisions didn't serious

1:11:09

injuries among both patients and medical staff Hospital officials were quick to respond to the incident shutting down the ISS

1:11:16

term and activating emergency protocols are there this is a tragic and completely

1:11:21

unacceptable event said the hospital's CEO the hospital are working with the

1:11:27

manufacturers of the technology to understand what went wrong the internet highlights the importance

1:11:33

of proper testing boy this is simply awful guys this this means from now I

1:11:38

will not be able to describe if an assignment is real or written by Chad

1:11:44

GPT it literally breaks my heart it's just awful I hope it blew your mind away

1:11:51

as much as it blew mine oh boy okay let's see that we have a few more

1:11:57

questions going on does chagibiti Integrated Medical electronic medical records I would say uh fortunately not

1:12:05

but I also have to say not for now I guess there will be vendors medical record vendors doing that very soon in

1:12:11

the future um can I use chat GPT to create a

1:12:20

personalized care pathway yes absolutely you can I I just so I received a private

1:12:25

email from a peer a colleague of mine who told me that he downloaded the data from his feedbit variable the Phoebe

1:12:32

device he uploaded it with some specific questions about fine-tuning his lifestyle habits and insights and it

1:12:40

received responses based on the data that that the question included so yes

1:12:46

it can provide uh care actually let me ask jaji pity a

1:12:52

very dumb very strange question I personally have a I'm a go back until I

1:12:58

describe it to you I two genetic tests have showed that I have um so-called

1:13:03

novelty seeking Behavior because I have two Gene variants that make that lead to

1:13:10

a serotonin reuptake deficiency which means even if I do something that that

1:13:15

makes me happy I will lose that um that feeling soon because I my my new

1:13:21

neural cells cannot reuptake serotonin thus I have to constantly look for new

1:13:26

things to to keep myself satisfied and dynamic and happy and that's called the novel The seeking Behavior it's also

1:13:33

called neophilia I'm very interested in chagibit and actually I have been trying to search

1:13:39

for this and other people's experiences with that and what they do about it but

1:13:44

uh what advice would you give to someone

1:13:50

who has a novelty seeking Behavior

1:13:55

what mental health advice would you give someone who hasn't nobody sick I am very

1:14:01

curious about it like I I did some searches about this on Google I found a few articles on psychology today and

1:14:07

stuff like that those contained a few bits and details but nothing extraordinary

1:14:13

again I hope it doesn't come up with anything useful uh it's important for them to understand

1:14:18

that why seeking out new and exciting experiences can be enjoyable it's also important to maintain about Jesus Christ

1:14:25

it's built it's perfect here are a few pieces of advice set boundaries to set limits on how much

1:14:32

time and money you spend on seeking out new experiences find healthy Outlets instead of confidence seeking out new

1:14:39

and potentially risky experiences I don't do that find healthy outlets for your desire for novelty learning a new

1:14:46

skill volunteering reflect on your motivations take some time to reflect on why you feel the need

1:14:53

Jesus God this is beautiful this is just extraordinary seek professional help thank you be mindful of your safety okay

1:15:00

don't you know I don't parachute it's a new thing this is spectacular

1:15:07

all right guys all right um a few more questions to go through if

1:15:13

you haven't lost now I'm just kidding of course we haven't lost hope in humanity these tools will will rise us it will

1:15:21

help us it will be the best assistant we've ever dreamed about and we will do

1:15:26

our job even better we will be forced if not motivated to constantly look for the

1:15:32

skills the things that make us exceptional human being because everything else everything that is

1:15:37

repetitive and or databased will be taken care of by these AI based tools

1:15:44

uh any legal issues have the approvals okay I think we've gone through all the major

1:15:51

things here and uh okay we are getting to the end so I'm

1:15:57

checking if you have any more questions we've had about I think 70 or 80 questions in the Stream I think that's

1:16:03

more than enough and it's quite clear based on how many you how many of you took part in this

1:16:10

stream and how many questions you sent to me before and um how many questions

1:16:17

um you asked during the stream that we will have to

1:16:23

um I have more streams about that I will have more live Q as about that it's just

1:16:28

simply extraordinary and I hope that you get why I'm so excited about this breakthrough Chad GPT alone not even

1:16:35

talking about multimet QA and all these medical chatbots just the fact that this tool is publicly and globally available

1:16:42

and you can ask all your silly questions and by the time uh throughout this time

1:16:47

you you build your own relationship with this technology I think it's

1:16:53

extraordinary so I hope that made this live q a i could answer some of the questions that buffer your minds I hope

1:17:00

that you saw how you could use them yourself and you will now go on chat GPT You Will Go On Me journey and and ask

1:17:06

really specific really exciting or just simple questions about your life your job your future and you will start

1:17:12

creating AI based images with the chat with the me journey and Dolly

1:17:19

um I hope that you enjoyed being here I'm I'm I cannot tell you how grateful I am for you for being in the chat for

1:17:24

more than like 75 minutes and asking all these really smart really thoughtful questions in the meantime of course as

1:17:31

usual we will keep on um analyzing Ai and all these tools and keep on publishing free articles

1:17:38

analysis every single day on the medical futurist um LinkedIn Twitter Instagram

1:17:44

uh Facebook and YouTube channels as well as on the medicalfuturist.com website if

1:17:50

you want to support us why we do that you can do this through uh three ways you can buy the ebooks we publish on

1:17:57

leanpop.com the medical futurist let me share just one book with you about the

1:18:03

future of AI if you want to get exclusive content I just but but that we publish on social media then you can

1:18:10

become a patreon a patreon for us on patreon.com the medical futurist and you

1:18:17

can also join the YouTube channel and support what we do so the team of 15 people at the medical futuristic and

1:18:23

keep on providing a huge amount of information and Analysis and details for

1:18:28

Fame in the meantime I hope that you will be safe and you will uh keep on experimenting with these AI tools and I

1:18:36

hope to see you at the next live q a where we have a very special guest I will describe and analyze the impact of

1:18:43

all the virtual mixed and augmented reality tools and Technologies and the

1:18:49

impact those tools can have on medicine and Healthcare so thank you so much again for being here meet me with us

1:18:55

today and have an excellent day bye

What is Artificial Intelligence and how does it work? What are the 3 types of AI? The 3 types of AI are: General AI: AI that can perform all of the intellectual tasks a human can. Currently, no form of AI can think abstractly or develop creative ideas in the same ways as humans.  Narrow AI: Narrow AI commonly includes visual recognition and natural language processing (NLP) technologies. It is a powerful tool for completing routine jobs based on common knowledge, such as playing music on demand via a voice-enabled device.  Broad AI: Broad AI typically relies on exclusive data sets associated with the business in question. It is generally considered the most useful AI category for a business. Business leaders will integrate a broad AI solution with a specific business process where enterprise-specific knowledge is required.  How can artificial intelligence be used in business? AI is providing new ways for humans to engage with machines, transitioning personnel from pure digital experiences to human-like natural interactions. This is called cognitive engagement.  AI is augmenting and improving how humans absorb and process information, often in real-time. This is called cognitive insights and knowledge management. Beyond process automation, AI is facilitating knowledge-intensive business decisions, mimicking complex human intelligence. This is called cognitive automation.  What are the different artificial intelligence technologies in business? Machine learning, deep learning, robotics, computer vision, cognitive computing, artificial general intelligence, natural language processing, and knowledge reasoning are some of the most common business applications of AI.  What is the difference between artificial intelligence and machine learning and deep learning? Artificial intelligence (AI) applies advanced analysis and logic-based techniques, including machine learning, to interpret events, support and automate decisions, and take actions.  Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.  Deep learning is a subset of machine learning in artificial intelligence (AI) that has networks capable of learning unsupervised from data that is unstructured or unlabeled.  What are the current and future capabilities of artificial intelligence? Current capabilities of AI include examples such as personal assistants (Siri, Alexa, Google Home), smart cars (Tesla), behavioral adaptation to improve the emotional intelligence of customer support representatives, using machine learning and predictive algorithms to improve the customer’s experience, transactional AI like that of Amazon, personalized content recommendations (Netflix), voice control, and learning thermostats.  Future capabilities of AI might probably include fully autonomous cars, precision farming, future air traffic controllers, future classrooms with ambient informatics, urban systems, smart cities and so on.  To know more about the scope of artificial intelligence in your business, please connect with our expert.

מהי בינה מלאכותית וכיצד היא פועלת? מהם 3 סוגי הבינה המלאכותית? שלושת סוגי הבינה המלאכותית הם: בינה מלאכותית כללית: בינה מלאכותית שיכולה לבצע את כל המשימות האינטלקטואליות שאדם יכול. נכון לעכשיו, שום צורה של AI לא יכולה לחשוב בצורה מופשטת או לפתח רעיונות יצירתיים באותן דרכים כמו בני אדם. בינה מלאכותית צרה: בינה מלאכותית צרה כוללת בדרך כלל טכנולוגיות זיהוי חזותי ועיבוד שפה טבעית (NLP). זהו כלי רב עוצמה להשלמת עבודות שגרתיות המבוססות על ידע נפוץ, כגון השמעת מוזיקה לפי דרישה באמצעות מכשיר התומך בקול. בינה מלאכותית רחבה: בינה מלאכותית רחבה מסתמכת בדרך כלל על מערכי נתונים בלעדיים הקשורים לעסק המדובר. זה נחשב בדרך כלל לקטגוריית הבינה המלאכותית השימושית ביותר עבור עסק. מנהיגים עסקיים ישלבו פתרון AI רחב עם תהליך עסקי ספציפי שבו נדרש ידע ספציפי לארגון. כיצד ניתן להשתמש בבינה מלאכותית בעסק? AI מספקת דרכים חדשות לבני אדם לעסוק במכונות, ומעבירה את הצוות מחוויות דיגיטליות טהורות לאינטראקציות טבעיות דמויות אדם. זה נקרא מעורבות קוגניטיבית. בינה מלאכותית מגדילה ומשפרת את האופן שבו בני אדם קולטים ומעבדים מידע, לעתים קרובות בזמן אמת. זה נקרא תובנות קוגניטיביות וניהול ידע. מעבר לאוטומציה של תהליכים, AI מאפשר החלטות עסקיות עתירות ידע, תוך חיקוי אינטליגנציה אנושית מורכבת. זה נקרא אוטומציה קוגניטיבית. מהן טכנולוגיות הבינה המלאכותית השונות בעסק? למידת מכונה, למידה עמוקה, רובוטיקה, ראייה ממוחשבת, מחשוב קוגניטיבי, בינה כללית מלאכותית, עיבוד שפה טבעית וחשיבת ידע הם חלק מהיישומים העסקיים הנפוצים ביותר של AI. מה ההבדל בין בינה מלאכותית ולמידת מכונה ולמידה עמוקה? בינה מלאכותית (AI) מיישמת ניתוח מתקדמות וטכניקות מבוססות לוגיקה, כולל למידת מכונה, כדי לפרש אירועים, לתמוך ולהפוך החלטות לאוטומטיות ולנקוט פעולות. למידת מכונה היא יישום של בינה מלאכותית (AI) המספק למערכות את היכולת ללמוד ולהשתפר מניסיון באופן אוטומטי מבלי להיות מתוכנתים במפורש. למידה עמוקה היא תת-קבוצה של למידת מכונה בבינה מלאכותית (AI) שיש לה רשתות המסוגלות ללמוד ללא פיקוח מנתונים שאינם מובנים או ללא תווית. מהן היכולות הנוכחיות והעתידיות של בינה מלאכותית? היכולות הנוכחיות של AI כוללות דוגמאות כמו עוזרים אישיים (Siri, Alexa, Google Home), מכוניות חכמות (Tesla), התאמה התנהגותית לשיפור האינטליגנציה הרגשית של נציגי תמיכת לקוחות, שימוש בלמידת מכונה ואלגוריתמים חזויים כדי לשפר את חווית הלקוח, עסקאות בינה מלאכותית כמו זו של אמזון, המלצות תוכן מותאמות אישית (Netflix), שליטה קולית ותרמוסטטים ללמידה. יכולות עתידיות של AI עשויות לכלול כנראה מכוניות אוטונומיות מלאות, חקלאות מדויקת, בקרי תעבורה אוויריים עתידיים, כיתות עתידיות עם אינפורמטיקה סביבתית, מערכות עירוניות, ערים חכמות וכן הלאה. כדי לדעת יותר על היקף הבינה המלאכותית בעסק שלך, אנא צור קשר עם המומחה שלנו.

Google Colab: Google Colab היא פלטפורמה מקוונת המאפשרת למשתמשים לשתף ולהריץ סקריפטים של Python בענן Graphics Processing Unit(GPU): GPU, או יחידת עיבוד גרפית, הוא סוג מיוחד של שבב מחשב שנועד להתמודד עם המורכבות חישובים הדרושים להצגת תמונות ווידאו במחשב או במכשיר אחר. זה כמו המוח של המערכת הגרפית של המחשב שלך, והוא ממש טוב לעשות הרבה מתמטיקה ממש מהר. GPUs משמשים סוגים רבים ושונים של מכשירים, כולל מחשבים, טלפונים וקונסולות משחקים. הם שימושיים במיוחד למשימות הדורשות כוח עיבוד רב, כמו משחקי וידאו, עיבוד גרפיקה תלת-ממדית או הפעלת אלגוריתמים של למידת מכונה. מודל שפה גדול (LLM): סוג של מודל למידת מכונה שאומן על כמות גדולה מאוד של נתוני טקסט ומסוגל ליצור טקסט בעל צליל טבעי. Machine Learning (ML): שיטה ללמד מחשבים ללמוד מנתונים, מבלי להיות מתוכנתים במפורש. עיבוד שפה טבעית (NLP): תת-תחום של AI המתמקד בהוראת מכונות להבין, לעבד וליצור שפה אנושית רשתות עצביות: סוג של אלגוריתם למידת מכונה המבוססת על המבנה והתפקוד של המוח. שדות קרינה עצביים (NeRF): שדות קרינה עצביים הם סוג של מודל למידה עמוקה שיכול לשמש למגוון משימות, כולל יצירת תמונה, זיהוי אובייקטים ופילוח. NeRFs שואבים השראה מהרעיון של שימוש ברשת עצבית למודל של זוהר תמונה, שהוא מדד לכמות האור שנפלט או מוחזר על ידי אובייקט. OpenAI: OpenAI הוא מכון מחקר המתמקד בפיתוח וקידום טכנולוגיות בינה מלאכותית שהן בטוחות, שקופות ומועילות לחברה. Overfitting: בעיה נפוצה בלמידת מכונה, שבה המודל מתפקד היטב בנתוני האימון אך גרועים בחדשים, בלתי נראים. נתונים. זה מתרחש כאשר המודל מורכב מדי ולמד יותר מדי פרטים מנתוני האימון, כך שהוא לא מכליל היטב. הנחיה: הנחיה היא פיסת טקסט המשמשת לתכנון מודל שפה גדול ולהנחות את הדור שלו Python: Python היא שפת תכנות פופולרית ברמה גבוהה הידועה בפשטות, בקריאות ובגמישות שלה (כלי AI רבים משתמשים בה) למידת חיזוק: סוג של למידת מכונה שבה המודל לומד על ידי ניסוי וטעייה, מקבל תגמולים או עונשים על מעשיו ומתאים את התנהגותו בהתאם. מחשוב מרחבי: מחשוב מרחבי הוא השימוש בטכנולוגיה כדי להוסיף מידע וחוויות דיגיטליות לעולם הפיזי. זה יכול לכלול דברים כמו מציאות רבודה, שבה מידע דיגיטלי מתווסף למה שאתה רואה בעולם האמיתי, או מציאות מדומה, שבה אתה יכול לשקוע במלואו בסביבה דיגיטלית. יש לו שימושים רבים ושונים, כמו בחינוך, בידור ועיצוב, והוא יכול לשנות את האופן שבו אנו מתקשרים עם העולם ואחד עם השני. דיפוזיה יציבה: דיפוזיה יציבה מייצרת תמונות אמנותיות מורכבות המבוססות על הנחיות טקסט. זהו מודל AI של סינתזת תמונות בקוד פתוח הזמין לכולם. ניתן להתקין את ה-Stable Diffusion באופן מקומי באמצעות קוד שנמצא ב-GitHub או שישנם מספר ממשקי משתמש מקוונים הממנפים גם מודלים של Stable Diffusion. למידה מפוקחת: סוג של למידת מכונה שבה נתוני האימון מסומנים והמודל מאומן לבצע תחזיות על סמך היחסים בין נתוני הקלט והתוויות המתאימות. למידה ללא פיקוח: סוג של למידת מכונה שבה נתוני האימון אינם מסומנים, והמודל מאומן למצוא דפוסים ויחסים בנתונים בעצמו. Webhook: Webhook הוא דרך של תוכנת מחשב אחת לשלוח הודעה או נתונים לתוכנית אחרת דרך האינטרנט בזמן אמת. זה עובד על ידי שליחת ההודעה או הנתונים לכתובת URL ספציפית, השייכת לתוכנית האחרת. Webhooks משמשים לעתים קרובות כדי להפוך תהליכים לאוטומטיים ולהקל על תוכניות שונות לתקשר ולעבוד יחד. הם כלי שימושי למפתחים שרוצים לבנות יישומים מותאמים אישית או ליצור אינטגרציות בין מערכות תוכנה שונות.