AI50 Connect

The Future is Now, Exploring AI, Automation, and Innovation in Healthcare and Business

July 08, 2023 Hanh Brown / Dr. Douglas Dew / Conor Grennan Season 4 Episode 181
AI50 Connect
The Future is Now, Exploring AI, Automation, and Innovation in Healthcare and Business
Show Notes Transcript

Don't miss our captivating event, "The Future is Now: AI, Automation, and Innovation in Healthcare and Business." Join Dr. Douglas Dew, CEO of Automated Clinical Guidelines, and Conor Grennan, Dean at NYU Stern School of Business, for an exploration of cutting-edge AI technologies.

Discover how AI is revolutionizing healthcare and business across various sectors. Learn about Dr. Dew's groundbreaking innovations and his unique system enabling machines to comprehend medicine through two-way conversations. Conor Grennan's initiatives at NYU Stern will be highlighted, fostering generative AI fluency.

Don't miss this thought-provoking event as industry pioneers share their experiences, inspiring stories of innovation and entrepreneurship. Explore the challenges and opportunities in AI regulations. Attendees will gain valuable insights into the transformative role of AI in healthcare and business.

Join us for an enriching event that expands your knowledge of AI, automation, and innovation in healthcare and business.

Find Dr. Douglas Dew on LinkedIn:
Find Conor Grennan on LinkedIn:

👉 See our Website:
🎙 Boomer Living Podcast:
👉 LinkedIn:

Hanh: 00:00:00
Hi, welcome. Thank you so much for tuning in to today's exciting LinkedIn live session. The future is now exploring AI automation and innovation in healthcare and business. Our focus today is a force that's reshaping every corner of the world. The powerful, promising and thought-provoking realm of artificial intelligence, no longer confined

Hanh: 00:00:23
in the realm of science fiction. AI has permeated mainstream sectors such as healthcare, education, finance, and transportation, alongside the tech hubs. So let's delve deeper into the transformative role AI plays across these sectors. In healthcare, AI's potential is boundless. It enhances diagnostic accuracy,

Hanh: 00:00:46
offers insightful interpretations of medical images, and paves the way for personalized patient care. Now, you might wonder what might be the goal. Well, health services tailored uniquely for every individual optimizing patient outcomes. Now, AI's influence is equally transformative in education.

Hanh: 00:01:10
It facilitates personalized learning experiences. Adjusting to each student's learning pace and style. AI's tools can also identify struggling students, ensuring that everyone receives the support they need when they need it. Now in the financial sector, AI is like a super fast, highly efficient worker. It can go through lots of data really

Hanh: 00:01:31
fast, which helps us make better decisions and spot any fraud or risk more readily, making things more secure for everyone. Now on the transportation front, AI steers the advancement towards autonomous vehicles promising, safer, more efficient journeys in reducing human error in traffic management. As AI takes the center stage, it also redefines the job market.

Hanh: 00:01:59
While it's true that automation may face out some roles, AI's evolution will undoubtedly spawn new careers in AI ethics, data management, AI system integration. Cybersecurity, AI specific law policy making, and much more. So AI isn't just replacing jobs, it's revolutionizing employment landscape, fostering innovation,

Hanh: 00:02:22
and paving new career pathways. So while discussing AI's potential, we must also consider its risk. It requires a careful balance between excitement and caution. AI's misuse, such as spreading DeepFakes or persuasive falsehood on a large scale can erode societal trust and distort truth. While this risk becomes especially

Hanh: 00:02:48
significant when considering the potential misinformation in medical advice or political discourse, privacy concerns are another potential pitfall. Current AI systems may lack transparency and unintentionally perpetuate societal biases. Leading to discrimination in areas like housing and employment. Well, these issues highlight the need

Hanh: 00:03:11
for rigorous oversight regulation and accountability among AI developers. So today we're excited to delve into these themes and much more with Dr. Douglas. Do the CEO of Automated Clinical Guidelines. And Conor Grennan, Dean of MBA and graduate students at NYU Stern School of Business.

Hanh: 00:03:34
We'll unpack AI's transformative role in healthcare and business. Demystify chat bot technologies like ChatGPT, and explore how these innovations are revolutionizing various sectors. So if you are passionate about AI, automation, business innovation, and the future of healthcare, you're in the right place. So welcome to the show, Doug and Conor.

Hanh: 00:03:59

Conor: 00:03:59
Thank you.

Hanh: 00:03:59
to have you both here with us.

Douglas: 00:04:02
And, good morning from Florida.

Conor: 00:04:04
Yeah, good morning.

Hanh: 00:04:05
How are you, you say from Florida? Where are you from, Doug?

Douglas: 00:04:08
We're in San Augustine.

Hanh: 00:04:10

Douglas: 00:04:10
And we have, uh, workers in Orlando as well. Great.

Hanh: 00:04:13
Great. How about you Conor?

Conor: 00:04:15
I'm in, uh, New Canaan, Connecticut, about 40 miles north of New York City where I, uh, commute to.

Hanh: 00:04:21
Awesome. Well thank you. Now, Conor, could you briefly tell us about your professional background and share an unexpected personal interests or hobbies?

Conor: 00:04:30
Sure, yeah. Thanks again for having me and great to see everybody in the chat and everywhere else. It's awesome to see everybody. So yeah, so my background is in a lot of different things, political science, and, uh, and I'm a writer by trade. I have, you know, I've, I've written bunch books and things like that.

Conor: 00:04:43
Uh, but more recently, you know, having gotten my a, uh, my b a from, uh, NYU and before that, uh, living abroad, I have a nonprofit in Nepal. I lived in Katmandu. I lived in Europe for about 10 years. But more recently, uh, having gotten my MBA and writing, uh, I came back to NYU Stern. I became the Dean of Students there.

Conor: 00:05:02
I've been there for about nine years, and more recently than that, my interest in AI is because, uh, my wife is in, uh, AI at McKinsey and Company. And so AI has been a big part of our family. And so at the advent of Generative AI, as you talked, well, not the advent sort of when ChatGPT came out, sort of like the more consumer facing product there,

Conor: 00:05:21
uh, I dove right in and then started, uh, Generative AI at Stern, which is, uh, the, uh, initiative that we have that's gonna, you know, training our, our MBA students and our faculty and everybody else. But, but I have a pretty diverse, uh, background leading up to this.

Hanh: 00:05:36
Thank you. Thank you for being here. All right. Could you do the same, Doug?

Douglas: 00:05:41
Uh, I'm an orthopedic surgeon. I'm still in practice. Uh, but I've been involved in a number of, uh, startups, uh, served on the board of one public company and. Actually in the late 1980s, uh, presented to the FDA, how to develop a software validation as a medical device, then went into practice and now working on things that help doctors.

Hanh: 00:06:03
Thank you so much for joining. We are here to share our ideas and also wanna hear yours and learn from you as well. Okay. Let's talk about AI. Now, I'm gonna pose a question to the audience. When you think of AI, what comes to mind? Is it robots thinking like humans or maybe

Hanh: 00:06:20
some real complex computer algorithms? So for those of you that are joining us, so please put in the comments, your thoughts. What comes to your mind when you think of AI? Well, robots thinking like humans and complex computer algorithms fall under AI, and that's where the heart of the debate lies.

Hanh: 00:06:44
AI in the simplest terms, is about machines. Simulating human intelligence. That means they can learn reason, solve problems, understand languages, and even recognize images or voices much like we do. But here's where things get a little tricky. Some folks argue for a machine to

Hanh: 00:07:02
qualify as AI, it needs to do more than just execute program test. Well, this should exhibit some form of self-awareness or consciousness, much like humans do now. But others believe that any machine learning system that can process information and make decision qualifies as AI and add. To add another layer to this

Hanh: 00:07:30
conversation, there's a whole narrow AI versus general AI debate. When we talk about narrow AI, we're referring to systems designed to carry out specific tasks. Like recognizing your voice or suggesting songs that you might like. Now this is the AI that we're most familiar with. Now, general AI.

Hanh: 00:07:55
On the other hand, it's like the holy grail of artificial intelligence. Imagine a machine that could understand, learn, adapt, and apply knowledge just like a human across a wide range of path. Well, this concept is very controversial and it's not too far from being fully achieved. So AI is a world full of diverse and sometimes conflicting ideas.

Hanh: 00:08:20
So Conor and Doug perhaps will ask Conor, so what is your personal take on the definition of artificial intelligence and do you think machines will ever achieve a level consciousness or self-awareness? And if so, What would that look like?

Conor: 00:08:36
Yeah, I think it's taking a turn that none of us really had anticipated when these large language models came out. So when we think about ChatGPT, that's what we're talking about. A large language model that essentially just predicts text, right? So it kind of predicts just the next word that's coming. But I don't get too caught up

Conor: 00:08:50
in that because this is not something that you have to have any technical background to use. That's the really startling thing to me. So when I think about how, like, so I train companies and I train people on how to use this, and I talk about it like an iPhone almost. You know, you don't have to sort of understand the underlying

Conor: 00:09:08
software of your iPhone. You're just ordering Ubers and Starbucks and booking flights and things like that. So when I think about where AI is going, I'm as surprised as anybody else, mostly because this has gone from something where it just seems to predict text and sort of sound like a human who really, really sounding like a human to the extent that you can't even really tell

Conor: 00:09:25
it apart from when you're speaking to a human or when you're speaking to ChatGPT. And in terms of what you were talking about about. AGI, Artificial General Intelligence, and whether something would ever become truly sentient. I think we are, I think this may be the model that we do it with. In fact, there was a research paper

Conor: 00:09:43
put out by Microsoft researchers, and this isn't, you know, people on Reddit just, you know, uh, speculating, but called sparks of general intelligence, where they realized that the difference between cert, for example, ChatGPT 3.5 and four was so alarming because when you put out, you know, put a problem to 3.5, for example, a slightly older model, and you took this thought experiment

Conor: 00:10:05
around, you had a laptop, a book, nine eggs, a bottle and a nail, and how do you balance them on top of each other? And Chat 3.5 was like, okay, put the nail and then the bottle, and then stack the eggs on top. It just didn't understand the concept of stability and everything else because it's a text-based linguistic model. But all of a sudden the chat, uh,

Conor: 00:10:22
4.0, which has just, you know, blown past 3.5 really, it started to understand the concepts of. Stability and going outside it's linguistic framework. And that's why some of these greatest researchers in the world are talking about sparks of this. So I actually think, and we'll get into this later about just the

Conor: 00:10:43
real power of this, that the large language model is going to be the key to actually becoming something like artificial general intelligence. And I don't think we're the, I mean, we're certainly years away, but I'm not sure that we're much further than that.

Hanh: 00:10:57
Yeah. Thank you. Thank you so much. So what do you think, Doug?

Douglas: 00:11:00
Well, I, I think it's in bits and pieces. And what I mean by that, uh, our work is actually in expert systems where you teach a machine to have a conversation. And in Generative AI, it's really a one-way conversation that can answer the question with very detail, with good details and good references that you can check.

Douglas: 00:11:21
But in a relatively short time, I think if you combine conversational systems with the large language models, You could certainly have the gamers working with Hol Hollywood to make an avatar, to pass a tour test in specific, uh, knowledge domains. They may know how to code it, might know how to beat Mario up at a video game. And of course, the, the worry is

Douglas: 00:11:42
it's, it's teaching, you know, from the water it's drinking. What happens if somebody poisons the water with misinformation, uh, to do that, let's say they break into an API and do good things or bad, I don't think it really knows what it's doing, but it can be trained to do certain things or answer certain things or carry on certain conversations

Douglas: 00:12:04
that can mimic that Turing test.

Hanh: 00:12:07
Mm-hmm. So now, Doug, so as a seasoned professional in healthcare it, in orthopedics, how do you perceive the role of AI in healthcare?

Douglas: 00:12:16
Well, I think it's, it's really two do different things that you're looking for. One is trying to gather information, uh, that's useful and valuable to the patient, to the doctor. And then the other part of it is all of the tasks that are, I call them menial, menial tasks that you know, you really have to do prior authorization.

Douglas: 00:12:37
You have to do clinical documentation. Those are the problems that need to be fixed in medicine because just look at, look at physician burnout. What does that really mean? Is all of the things we have to put up with in the office, in the hospital, um, that don't contribute to patient care. That's where I see the biggest difference. And what I mean by that, let's say it's

Douglas: 00:12:57
annoyance, AI, some pop up in the EMR quality measures that coming from c m s that really don't move the needle. A prior authorization where sometimes third parties infuse chaos that make money off the chaos. Um, you imagine in Florida, we have to deal with every Blue Cross in the country to do prior auth, but we don't know what their policies are.

Douglas: 00:13:22
And then billing and coding the headaches with that. And the EMRs themselves. These are built left over from the 20th century, and you talk to the CIO everything is fine, but they say the doctors need more training. No, we need a, an intelligence system, or at least intelligence built into the EMR.

Hanh: 00:13:39
Mm-hmm. Mm-hmm. Great. Now, can you share about a patient care innovation that was significantly enhanced by the use of AI?

Douglas: 00:13:49
Well, I, I'll take, uh, I'll take, I'll, I'll put it, uh, put it in a very different way. Our American category, orthopedic surgeons used an AI algorithm in 2016 when it parsed out ICD 10, and it noticed the Z codes. So when CMS asked for quality measures, we recommended that you use the Z codes for communication

Douglas: 00:14:10
and understanding the patient. They're not used for billing. Our medical director said, Doug, CMS thinks this is a good idea. Uh, it was developed by CDC and they didn't know it existed. Now, seven years later, social determinants of health and that type of coding is important. So that's something that's

Douglas: 00:14:28
happened, but it took seven years to really get to that point. And in terms of prior authorization, uh, we use a system that's been developed in stealth where the people have been using it are now at six years with no prior auth turndowns and no peer-to-peer reviews. That gets rid of a big headache within medicine.

Hanh: 00:14:50
Mm-hmm. Mm-hmm. So now, in your work towards price transparency, how can AI contribute to the win-win situation for both patients and providers?

Douglas: 00:15:02
Well, if you look at, I'm, I'm an orthopedic surgeon that does knee surgery. No one realizes what the cost of the knee surgery is. So we had, uh, some people ask us, well, could you put together all the prices of drugs? Because we don't know what drugs are $10 a month and what drugs are $400 a month? Plus we have to get 'em approved for

Douglas: 00:15:20
prior auth, or we have to use specific formularies from different payers. Uh, this last year our group won the National PBM Innovation Award because we took PRI pharmacy transparency for all drugs and by class, by price, and also linked consumer information for education and like make clinical links to the manufacturer's website. Uh, and this means for drugs that require

Douglas: 00:15:44
prior auth dynamic questionnaires are provided that writes the clinicians note. At the same time, if you're coming from pharma and there's a new drug coming out, we can quickly edu educate the clinicians on appropriate use criteria cost in a transparent way, typically next business day based on our technology. And this means you can put evidence-based guidelines in the EMRs typically

Douglas: 00:16:10
within the next business day. This, this is a game changer.

Hanh: 00:16:14
Mm-hmm. Now, Can you share your experience in aligning innovative, possibly AI-driven solutions with the FDA regulatory compliance?

Douglas: 00:16:24
Well, I came from a FDA background long ago, and I served as an entrepreneur resident at the fda A, so I think regulations are important, but you also don't wanna put regulations that get in the way. So if you look at AI, I think the best way to think of it is it has to be transparent. It has to be something

Douglas: 00:16:43
that everybody understands. And I think it has to have, if you want to call it a prime directive or mission statement, to really understand what it's doing. And what I mean by that, you have to have a mission statement. Are you doing the right thing for the patient? That's good.

Douglas: 00:16:59
Are you doing it to make more money? No, that's not good. Are you making more money for the insurance of the pbm? That's not good. Do you deny, uh, car as a success metric? That's not right. But I think when you go to AI and regulation, You're going to have to go and have regulation locally through

Douglas: 00:17:15
institutional review boards because many of the people touting it are really self-serving and self-promoting. And you really need somebody to stand back locally and look at an institutional review board if it can affect patient care nationally. If it influences the, the physician's, uh, decision on treatment, I think it needs to go to the FDA.

Douglas: 00:17:40
And then there needs to be groundwork on security that the system can't be built to break into an API can't be built to give misinformation to generative systems. I mean, that's just as bad. It might feed on the information that's on the web. What if it's misinformation? What was the original source? You have to understand

Douglas: 00:17:59
the original source. That's probably the most important thing.

Hanh: 00:18:04
Mm-hmm. Mm-hmm. So, how has a combination of medical lasers and AI resulted in improving patient outcomes through your experience?

Douglas: 00:18:15
Well, in orthopedics, not that there are not that many applications of orthopedic, uh, clinical use. And what I mean by that, where we un would had researched in the late 1980s, we understood how to develop software as a medical device. I understood clinical trials running IDE clinical trials, 510(k)s, and one PMA. But that information needs to be

Douglas: 00:18:38
documented where the biggest, uh, difference in medicine and orthopedics with lasers, uh, we actually put markers on implants to track those implants. So all of those ID numbers are laser engraved. Uh, and I, I'll work with that company, um, many years ago. So that's been the biggest impact in orthopedics is the

Douglas: 00:19:02
marking of implants for tracking. And that was kind of my introduction, informatics.

Hanh: 00:19:08
Okay. All right. I wanted to acknowledge a few of the folks in the audience here. Again, we thank you for your time. There is Rene, Lila, Julia, Barbara, thank you so much for joining us and thank you Jordan and Chris Gailey. If you have any questions, please feel free to chime in.

Hanh: 00:19:31
Again, we appreciate your attention and of course let us know what your thoughts on this very important topic that we're discussing. So Doug, so how can AI streamline the alignment of clinical guidelines with the FDA compliance at your company, Automated Clinical Guidelines?

Douglas: 00:19:49
Well, we have a different type of AI that not many people are talking about right now. If you go back to the 1990s, there was lots of excitement and hype about expert systems. The problem was, is they really couldn't be scaled in terms of, of an, except for a niche product. And the cost of updating them

Douglas: 00:20:07
is, was cost prohibitive. Well, those expert systems, we've developed a technology where the algorithms are transparent. They can be read by a physician, they can be read by the fda, and they can be read by a machine. And so with that case, if we're doing diagnostic and treatment work, uh, we, we, we've really built the system

Douglas: 00:20:28
as a FDA medical device for software. It's a medical, as a, uh, device, and validated that software over the last few years. But that's only for diagnosis and treatment that would influence the physician's decision. For prior authorization. That's really about medical policy. It's not really influencing the decision,

Douglas: 00:20:48
uh, other than turning the physician down and denying care for the patient. But it's good also at utilization management for doing the right thing to the patient. You don't want overutilization, you want the right thing, the right medicine, the right tests, the right surgery when it's appropriate. Uh, so what we've done is actually

Douglas: 00:21:06
develop software that's now has 10,000 algorithms and those 10,000 algorithms can be kept up in real time. Our machine learns every day. And what I mean by that, because we can put up an algorithm typically next business day, uh, the cost has gone way down. If we had to send out 10,000 algorithms, uh, to India at, uh, let's say 18 months

Douglas: 00:21:34
to put those together, if we had 10,000 algorithms, an outside group calculated, it would take 3,400 contract programmers. Working five years at a cost of 1.7 billion to write that amount of code. And that's why expert systems really hadn't caught on, uh, that, that's our magic, where we can write zero defect code, sec code sets in seconds. And what I mean by that, we even wrote

Douglas: 00:21:59
a 10 million line algorithm that's updated annually with almost no debugging required because it's completely machine written with almost zero defects. So that's, that's kind of a technology that can really make that information useful, valuable, bring in the transparency, bring in the prices, and really present a rational economic sy uh, system.

Douglas: 00:22:23
Uh, as a knee surgeon, I don't know what the cost of a knee surgery is until somebody gathers that information up.

Hanh: 00:22:30
Mm-hmm. Mm-hmm. That's great. Now, feel free both of you, look at the, um, the comments and the feed here, and feel free to see if there are any questions. I feel like we owe it to the audience to answer if it pertains here. Hello everybody.

Conor: 00:22:47
Yeah, I've had a chance to just, uh, watch it scroll. This is a phenomenal audience. I don't know how you gathered all these people here today. They're, they're amazing. They're just like super thoughtful and everybody wants to connect. And, uh, I know Doug is on, uh, LinkedIn. I'm at Conor Grennan at LinkedIn.

Conor: 00:23:02
We have this phenomenal community there, so please come join and follow us because, uh, this is super cool. Uh, but yeah, I think a lot of people are asking, you know, about Generative AI and sort of, uh, you know, and how all this figure is in and everything else. But, uh, but yeah, this audience is awesome.

Hanh: 00:23:18
Yeah, I thank you. I thank you so much for joining. So yeah, feel free to, uh, ask specific questions and, uh, greetings to you all. So I'm gonna go to Conor. I have a couple questions regarding education. Okay. So how do you think Generative AI contributed to the creative

Hanh: 00:23:41
immersive, interactive learning experiences at NYU Stern?

Conor: 00:23:46
Yeah. Well, a few months ago, you know, we didn't know anything about it. Like, like all of you. I mean, except for the great machine learners out there and everything else. I mean, on learning, let me say this, you know, we have sort of our typical education. Like if you all kids out there

Conor: 00:23:57
and you're wondering how they're ever gonna learn to write ever again, for example, that's tricky. And then there's also the learning as it comes, you know, for the sort of, it's like for the corporate side as well. So at in education, at Stern, we sort of sit at the middle, right? So I'm in an MBA program and these are students that aren't kids anymore, right?

Conor: 00:24:18
They are, you know, Fully formed adult in their late twenties, depending on your, not fully formed adult, I guess, but also, you know, people who are about to go out and change the world. And so what, what we determined was, you know, kind of like looking at how this is all happening. I was like, okay, so we need to actually like really level these folks up because

Conor: 00:24:37
I believe, and think of this as a statistical thought experiment, like with a thousand monkeys and a thousand typewriters and a thousand years, and you can write the works of Shakespeare. I believe that if I could train everybody on ChatGPT, and on Generative AI really well, we could actually solve education. I know that sounds huge, but I really believe, uh, we can do that because

Conor: 00:24:56
ChatGPT allows you to do a few different things that are really transformational. So lemme just take, you know, the education of, again, if you guys have kids or anything like that, you'll know that all of a sudden, you know, kids, instead of writing a five page paper on Hamlet can just say, Hey, ChatGPT write me a five page paper on Hamlet and it will do it. And so this is terrifying for teachers

Conor: 00:25:13
of course, and parents too, who want their kids to know how to do this stuff. So basically, like number one, we have to kind of think, okay, so how do we figure out like new proxies for grading? Meaning like, how do we figure out how to test kids on their knowledge? And I think we're actually a really exciting time. Maybe writing a five page paper

Conor: 00:25:30
on Alet is not the best way to teach kids about a Hamlet. Maybe it's, hey, you know, go back and like, you know, learn with ChatGPT and then come into the classroom and discuss. So on that front, on the education front, what's so interesting about that to me is the, just the sheer experiential learning that we can do. So I have a, I have a YouTube channel,

Conor: 00:25:49
which uh, just sort of like does like a lot of fun stuff around ChatGPT and just, you know, uh, a very recent one I was talking about, you know, this specific topic because I have a 14 year old son and he's like, oh, I have all this stuff on, you know, education on, you know, German immigrants coming to New York in 1885. And he is like, it's kind of dry. You know, if you've ever read

Conor: 00:26:07
about that, like, and I see we have some people from Germany. Super interesting, but also the way it's written about it dry. So all of a sudden he created a German immigrant that said, tell me what you're seeing. What are the sights and sounds around you? Where are you going right now? Where are you singing at?

Conor: 00:26:22
And oh, the German immigrant in 1885 going around New York City and you know, parts of the cell. I always thought parts of the cell would be really, really fascinating. So, but not in a textbook. So I turned them into marble characters and I had them just do this epic battle with like the villains of viruses and things like that.

Conor: 00:26:37
So I think in terms of education, that's where we start. We start with just making it more interesting and more experiential. And then, you know, in the NBA side, going in into the corporate side, this is a phenomenal learning tool and we can talk about that more. But ChatGPT helps people learn because humans don't tend to be

Conor: 00:26:53
very good learners because we're too worried about our status to ask dumb questions, but ChatGPT we can do that. So I think it's totally transformational.

Douglas: 00:26:59

Hanh: 00:27:00
So do you see a future where, let's say elementary, starting at elementary, all the way to college and so forth, somehow we got a package. Generative AI, ChatGPT, into their learning experience as early as possible and actually encourage it, but also still encourage critical thinking, creativity. Productivity. Right?

Hanh: 00:27:23
Because I, I see, and I'm sure many of you do that it's still perceived to be cheating. Not necessarily. I see it as your tutor, your coach. It's almost like it's a another world of learning that perhaps you didn't have right next to you. So hopefully in the education system, the sooner, the better that we can integrate

Hanh: 00:27:46
this into the younger generation. To, to use it productively and also continue to cultivate their critical thinking skills and so forth. Because I think one doesn't replace the other. It doesn't have to.

Conor: 00:27:58
Yeah, I, I totally agree. I think that, in fact, this forces us into critical thinking in a way that we've never had to be forced into before. Because if you think about like how education's been the same for, I don't know, 500 years or a thousand years or something, it's like, Hey, here's this question. Go answer this question, you know,

Conor: 00:28:13
memorize the answer, then come back and tell the teacher, and the teacher will give you a grade. That's not a great way to actually learn, right? So we can think about this on the elementary level and you know, sort of young kids and how they learn and instead, you know what I sort of, I have, you know, on my, I do a ton of LinkedIn

Conor: 00:28:29
posts around, um, and again, I'm not con redin, but like, you know, prompts and how to do this, especially with education. But one of the things is, Listen, like if my daughter has to teach, you know, has to learn about the Great Wall of China, why, you know, the professor could, or the teacher rather, could send her back and say, okay, next class you're gonna teach about the Great Wall of China

Conor: 00:28:45
and you're gonna sort of get ChatGPT to come up with this phenomenal lesson plan, but actually a fun lesson plan. And who's better at that than kids to come up with creative ways of thinking. And so, one of the things I do a ton as well as, you know, as well as being, uh, at NYU Stern, I go around and I do a ton of training with large corporations, the C-suites and, and leadership at

Conor: 00:29:04
really, really large corporations. And I find that they get held up in the learning much more than kids because in the same way kids can learn languages because they don't have these inhibitions about if they're gonna sound stupid. I'll tell you what, like C-Suite people and, and even you know, us, leaders on this call and all you folks, uh, who are out there commenting and everything like

Conor: 00:29:23
that, you are leaders and we're leaders. But here's the problem. We get caught up with, you know, again, are we gonna look dumb to people and everything like that? Kids don't care about that. So here's the phenomenal thing about ChatGPT. As I sort of alluded to earlier, humans are terrible learners.

Conor: 00:29:39
We're terrible learners. Because if I was gonna ask somebody a question about what an API is and they start explaining it to me, I'll say, oh, well you know how the internet works. I'll be like, yeah, I know how the internet works. And then they'll say, okay, well imagine that. You know, and then they'll

Conor: 00:29:50
start explaining it. But the fact is, at the end, I'll know a little bit more. But if you gave me Truth Serum, I would've said, actually, I have no idea how the internet works. And they'll say, oh, well it works in protocols. I'll be like, I actually don't know. And at some point they're gonna be

Conor: 00:30:01
like, Conor is an idiot, and they're never gonna talk to Conor Grennan again because they're gonna think this guy asked a lot of dumb questions. Right? So my relationship with that person is more important than learning that information. ChatGPT does not have that judgmental bone in its body.

Conor: 00:30:16
So in fact, what kids do really well is they're like, Actually, I don't get that, actually. I don't get that. And then can you explain it in an analogy? Using ballooned animals and then like, and then make this into a, you know, make the parts of the cell into a circus and teach me that way. But adults don't yet know how to do that.

Conor: 00:30:31
So what I would say when I'm sort of teaching big leadership teams that some of the biggest companies in the world is first, learn how to learn. Remember that you don't have to stop asking questions. This thing will actually teach you. So I think that is actually the future of education, learning how to learn first.

Hanh: 00:30:46
Mm-hmm. You know, I echo that because I think what you, let's say put in is what you put out. You're prompt. Engineering is everything. However, let's say you're not pleased with the outcome is typically because your prompt engineering may not be as precise as it needs to be. So what that does, at least to

Hanh: 00:31:02
me, is that it triggers me to think harder, be more creative. Right. And I don't necessarily think that it's hallucinating. And, and sometimes it does. It triggered me to improve myself and to learn to. Create better prompts. So I think in so many ways,

Hanh: 00:31:20
other than productivity, it's helped me to be more creative. Alright. So can you share how Generative AI has customized educational content for your courses?

Conor: 00:31:34
Yeah, I mean, listen, this is still an open field, you know, I mean, how I think every professor is gonna have to come up with their own ways. I've been helping them, you know, having, you know, imagine it's sort of like during Covid, like imagine if you were always online, like how would've you have developed this course? So I have to go in with professors

Conor: 00:31:48
and think, look, let's ask ChatGPT we're actually ask the very, the very tool how to, how to do that. But I think also to your, to your earlier point about proms, listen, ChatGPT only reflects your experience. And so if you are getting something that you are not satisfied with, the problem is not the tool, the problem. And if you know some American standardized

Conor: 00:32:08
test ChatGPT four, I mean it, ACE is the lsat, which is the, uh, the legal, you know, the, the law exam and the mcat, which is the medical exam. Uh, tests and the, uh, you know, the gmat, which is the business school exam. I know we use a lot of acronyms and so I'm trying to spell out what those things are, but if ChatGPT four can do all that, then it can certainly solve your

Conor: 00:32:31
problem about how to create a performance management system for your small company. So if it's not, this is the big thing. We tend to have get stuck. Our brain gets stuck a little bit, our brain gets stuck because we're so used to Google and asking for something and getting back response. Han, as you just said, ChatGPT is for, is not like that.

Conor: 00:32:50
This reflects your experience. You have to go back in and iterate and iterate and iterate. And whether that's, you know, developing a syllabus for professors or just solving your day-to-day problems or kids working on a paper, continue to have a conversation and treat it like a human.

Hanh: 00:33:03
Mm-hmm. Mm-hmm. That's great. So Doug, what do you think?

Douglas: 00:33:07
I find it very interesting, the comment of cheating.

Hanh: 00:33:11

Douglas: 00:33:11
Uh, we've developed algorithms for pretty much every diagnosis in the emergency room. Which means you come in, what should be done in the ambulance, what should be done when they hit the door, what labs to do, what tests to do, when they should be admitted, when they should be discharged. Are they an ops day? Are they a full admission?

Douglas: 00:33:29
Are they, uh, going to be going home? And what should be done? And when I gave those algorithms to some medical students, they said, this is like cheating those exact words. And I think if you go back when I took the uh, SAT I was not allowed to have a calculator that was cheating. And I think as long as we consider it a calculator and a modern tool,

Douglas: 00:33:50
it's not like saying we can't get on an airplane and, and fly. Uh, it's saying that the memorization where most medical doctors have been chosen for the memorization skills are not gonna be as important. It's gonna be important to teach logic, reasoning, and terminology. And by doing that, AI can be, whatever you define as AI can be your peripheral brain.

Douglas: 00:34:15
And as long as you can carry a conversation on that and it knows the next question, not just the next word, that will be the game changer. It has to be cognitive.

Conor: 00:34:25
Can I just, can I just second what Doug is saying? I hope everybody really heard what he was saying there cuz it is so critical, right? So I think what Doug is saying there is. It's even at the professional level, like he's dealing with real professionals here there's this sense of like, well, we're cheating. Why is that?

Conor: 00:34:38
It's because as kids we are taught that if we are getting answers from somewhere else outside of our own brain or asking your teacher that it's cheating, we have this drilled into us in the same way that, you know, we're scared of closets at night or something like that. Right? But I think what Doug is saying here is so, so critical.

Conor: 00:34:54
And again, Doug do on LinkedIn and Conrad on LinkedIn, this is where you go, folks, this is what's so important about what he's just saying. This is as he's saying, almost like a a, a helper brain. And it's not cheating cuz it's just reflecting your own experience. This is where you would go for, you know, if you're sitting

Conor: 00:35:10
around, brainstorm right? With two other people, invite ChatGPT into that conversation. It is the most phenomenal brainstorm. But also what Douglas was saying was around menial tasks. Let this thing do your menial tasks. There is nothing to say that being, you know, a doctor or an educator or a marketer or anything else requires you

Conor: 00:35:26
to keep doing all your menial tasks. Figure out how not to do that. And I, you know, I, again, I teach this whole framework around this kind of thing, but just, you know, there's an efficiency play here where if you just say, Hey, ChatGPT, write this email to my boss and explain why I need six more people working for me. ChatGPT will do that

Conor: 00:35:43
faster than you can think. And it doesn't even know who your boss is or who you are or what you're working for or why you need six people. And it will craft the most beautiful and compelling email. It's surreal. And the more information, ha, as you said, the more information you give it, the better it's gonna go.

Conor: 00:36:00
But I really encourage you to follow Doug's advice. This is not cheating. All this is doing is helping you kind of free up your prefrontal cortex to actually do critical thinking. And then like, let's, let's make subservient those menial tasks, let's outsource them to this tool.

Hanh: 00:36:13
Absolutely. You know, how can you cheat, like for yourself, because I feel like when I do use it, Properly, it's actually improving my overall, not only productivity, but my creativity. You know, I'm starting to think bigger. Like before there's certain tasks that, you know, I, I just don't wanna deal with. It's, it's just either repetitive or

Hanh: 00:36:32
perhaps you don't have the skills. Suddenly you can overcome that and then you end up spending the time to nurture your creativity right. To the back, to the soul of your work of what you're meant to do. So, yeah, I, I think we have to kind of change that paradigm that it's not cheating.

Conor: 00:36:53
Yeah. And esp I love that you say creativity. I think that, you know, you go back to the original question of where do we think AI was going? And I think we thought, you know, the people that will be safe will be like the creatives and everything else. I'm not saying this thing is human or it is creative, but it

Conor: 00:37:06
acts human and it acts creative. And even if you just throw this out for brainstorming and whether it's in, you know, in duds field, in in medicine, or in, in my field in training and everything else, This is a tool by which you can get people thinking. And when I train on, uh, you know, these executive teams, I train on learning development teams.

Conor: 00:37:23
I train on marketing teams. I train, uh, C-suite and executive teams again at some of these largest companies in the world. Here's the thing, ha. And you just mentioned try to be creative first because people get wa locked into their way of working and it's like, well, it has to follow this pattern. This is why kids are so good at it.

Conor: 00:37:39
Kids are like, whatever. Let's just try to sort of, you know, see if you can write a poem about vampire butterflies on the moon. It, they don't care. And that creativity really sparks some incredible prodit.

Hanh: 00:37:47
Mm-hmm. That's great. All right. So how are you both planning to further integrate Generative AI into, let's say, business education or medicine? So, Doug, do you have any ideas?

Douglas: 00:38:00
Well, I think Generative AI is good at writing, uh, letters, appeal letters, as an example. But you have to be careful what happens if the x-ray report on a knee before a knee surgery for prior auth has the same answer for every orthopedic surgeon in the country. That's fraud. So you have to look at your sources.

Douglas: 00:38:19
It has to be individualized to what you're really doing. And again, check the sources. I think that's the biggest problem and the biggest advantage of it. I mean, I, I have a son who's a producer in gaming and they look at it completely different. They look at it is, it's gonna help the coders.

Douglas: 00:38:38
It's gonna, it's really going to have a lot of coders lose their jobs. Um, and they, you know, we talk about APIs and medicine. They've been doing it for decades and, you know, we're kind of new at it. When you give that as an example, you know, many of the people, uh, Talk about APIs as being so important and this, that, and the other.

Douglas: 00:38:57
Um, you know, I was never good at poetry at, um, in medical in, um, undergrad I hated poetry cuz I didn't want to do it. It wasn't my interest. But just before this meeting, I went on ChatGPT and I said, can you write a poem about expert systems and what they can do? And I'll, I'll read it real quick. I think it's important. So it says, um, an expert system,

Douglas: 00:39:20
so smart and wise with knowledge and logic, it never lies. It processes data with speed and grace and provides solutions in every case. Its algorithms are sharp and true. It's reasoning are always sound and new. It's an expert in its field, no doubt a system we cannot do without. So here's to the expert system. So grant a normal of

Douglas: 00:39:42
technology at our command. Now it, it explained an expert system based on papers that were written in the nineties. When you look at that information, but yet it put it, put it into a poem, it put into some, some, uh, a form that everybody could understand and I wouldn't be good at it. That's not my expertise.

Douglas: 00:40:06
Now, the poets and the audience, they may not like that. Uh, it's the same thing as the calculators where the people that were good in school at doing, uh, multiplication tables and the subtraction division in their head, that's no longer important. You go on your phone. So I think we have to look at things of what's important, what's needs

Douglas: 00:40:27
to be learned every day, what's not going to be important in the future. Uh, I look at prior au gonna be gone in two years, less than two years. Uh, it's, it's awful when the, the patient and the physician know they're gonna get approved before the payer ever looks at it. Uh, and that's, that's to me, is the biggest breakthrough in medicine.

Douglas: 00:40:48
It should have gone away long ago, but. Not even just these 300 policies for a payer, it can now apply to all 70,000 diagnostic codes. So you can actually scale that utilization management to a billion office visits, 70,000 different diagnostic codes and treatments, and suddenly the memorization goes away. That's important.

Douglas: 00:41:12
And then you really have to understand, again, the logic why you're doing it and concentrate back on the patient, not, you know, doing a bunch of crazy paperwork to get paid or get the patient treated.

Hanh: 00:41:24
So true. So true. So what's your take, Conor?

Conor: 00:41:28
Yeah, I mean, there's so many directions to take it. I would say that, uh, I guess when I, again, when I do trainings on this, at these, at the, at the larger corporations and everything like that, I really try to steer people away from thinking about Generative AI and ChatGPT as just a tool. Meaning, you know, a lot of folks are just saying, well, you know, we wanna,

Conor: 00:41:49
you know, use this and talk toward data and get, don't get me wrong. Like, that is a needle mover. That is a profound impact on industry, how you can take your own data and, you know, analyze it and talk to it in natural language for yourself internally, for clients and everything else. But what I talk about, and again, on, on on LinkedIn, I do a lot of

Conor: 00:42:07
this around, you know, marketing and learning development and, and HR and everything else, is, this is an a, a chat, I'll say Generative AI and ChatGPT specifically is an augmentation of your own experience and your own intelligence. And that is why it is not a tool. Google is a tool. Google, and you know, and Salesforce is a tool.

Conor: 00:42:27
Salesforce is not an existential threat to humanity though. I mean, this is the power of this thing that we're talking about. And so what I talk about. What I'm dealing with teams is tell me first of all, what are the biggest three things that take up your day and let's see if we can't condense them. So in learning and development,

Conor: 00:42:41
they're building a lot of curriculums or you know, training workshops. And we can condense that from 60% of your day down to 10% of your day. In, in hr it's job descriptions and building those things out. We can take that from 50% of your day down to 15% of your day. All these things allow you to actually use your brain to

Conor: 00:43:00
do really interesting things. Marketing, it's just coming up with ideas. This can condense that from 70% of your day down to 30% of your day. It's an augmentation of everybody's experience. And we just saw it with Doug and the poem. I was reading the chat and kind of like laughing on the chat cuz some people are like, wow, that's really cool.

Conor: 00:43:16
Some people are like, that's a terrible poem. That's the point. People, that is the point. It augments what you are. So for example, if I, I am also not a poet, uh, Doug, but like, but if I was trying to sort of figure something out, like. Write a poem to me,

Conor: 00:43:28
that's a pretty good poem. Or if I'm trying to figure out how to write, you know, how to do a marketing campaign for a health and beauty company in London, which I helped them create, I'm gonna get like two levels deep into ChatGPT and be like, wow, this is a really interesting information. If you are a chief marketing officer, a cmo, and a health and beauty company,

Conor: 00:43:47
you'll be at the level I'm at and be like, this is really elementary rudimentary. This is not interesting because it's reflecting your own experience and not your intelligence, but your intellectual curiosity. So you'll have to go deeper and deeper into ChatGPT to do that. So if you are a poet and you're reading this poem, you're gonna

Conor: 00:44:02
be have, you're gonna say like, now write me an EE Cummings. Now change this. Take out that language book. You are gonna go deeper. Whereas me and Doug are like, yeah, that's a great poem. Check that out. You know, because we are an experienced poet, it reflects you.

Conor: 00:44:15
So don't let somebody else's experience with Chachi B team Generative AI define your experience. You have to find your own use cases, and this is what I train people on all the time.

Hanh: 00:44:23
Yeah, I echo that. For instance, I'm not a poet, but if I were to get the results the same as Doug did, if I didn't like it, like you said, I will go deeper and deeper. I would tell it to provide me different iterations to capture the output that I'm looking for. Okay. So now every time that I do this,

Hanh: 00:44:41
it requires critical thinking and it requires me to think about. What is it, the ideal output that I want? And then I also insert my own insights to it. So, to me, like I said, you know, it, it gives you what you, you provide. You're, it's all in your prompt engineering, the better, concise and so forth.

Hanh: 00:45:05
You're more likely to get better results. Now, could you share how Generative AI could be supported with your mission with the Next Generation Nepal?

Conor: 00:45:16
Yeah. Thanks for asking. I'll, I'll be, I'll be quick on this, but I do, my passion is, uh, Nepal, right? So about 15 years ago I was out in Nepal and I started an organization called Next Generation Nepal, uh, which rescues traffic children and reunites them with their families. Uh, and so that is, that is my passion.

Conor: 00:45:32
By the way. If you ever see a photo of me getting an award from the Dalai Lama folks, that is real. It's as crazy to me as it will be to you, trust me. But that's not a fake, I don't want people to say it's a fake, it's not a fake. But anyway, so we're out in Nepal and we're doing all this, and when

Conor: 00:45:46
Generative AI came out, I started thinking, you know, we have this moment in time, so often it's, you know, the rich getting richer, the poor getting poorer, the digital divides, you know, dividing us and everything else. And we were saying earlier, uh, today in Han you were talking about this as well, this is the thing. We have an opportunity to really

Conor: 00:46:01
bring people together here. We have an opportunity if, if you come from, you know, are living in a wealthy community and it's, you know, you grew up and maybe were able to have a tutor or something like that. I'm sure that there's a ton of people like me growing up that we couldn't afford a tutor growing up. ChatGPT is a phenomenal tutor.

Conor: 00:46:20
I'm thinking about the actual use cases for people, you know, the, the young kids living in Nepal and everything else, who all of a sudden have at their disposal this incredible tutor both in language, you know, physics, medicine, everything else. So that's one huge thing in Nepal. The other thing in Nepal, and I go out to some of the most remote

Conor: 00:46:37
parts of the world, uh, you know, in the, in the Himalayas. So the. These are places where literally there's no roads. They've never seen a wheel wheeled a vehicle in their entire life. And I walk from village to village finding families of parents. This is next generation De Paul's work

Conor: 00:46:49
so that when I get out there and I've seen, I used to take out a digital camera and I would do this and they wouldn't even know what I was doing. I would show them the back and they would look at their face and they wouldn't recognize their own face because they'd never had a mirror in their life. This is the kind of remote place I'm talking about.

Conor: 00:47:04
But Han, when you think about what this can do, and this is Doug's realm too, I'd love to hear Doug's thoughts on this because now all of a sudden it's not perfect medical advice. But if you can give them access to places where doctors cannot go and say, Hey, these are my symptoms, and forget about when it becomes multimodal. Even right now, these

Conor: 00:47:20
are some of my symptoms. What can I do and have a doctor, and I'm using that term lightly, Doug, cuz Doug has worked very hard to be, get that, get that affiliation. But when you can put this in the hands of people who do not have medical attention, to me, that is a game changer in, uh, Generative AI. It's a real matter of life and

Conor: 00:47:36
death out in a place like Nepal.

Hanh: 00:47:39
Mm-hmm. What's your take?

Douglas: 00:47:40
I'll, I'll comment on that. We actually, uh, have designed the system for other countries as well. Hmm. When the calculator of medicine, you go in and look at all of medicine, it's de-identified data, so therefore it can cross borders. But it also has Google language, uh, Google Translate audit.

Douglas: 00:48:01
So it's in over a hundred different languages. So if you're going into Puerto Rico as an example, everybody's trained in English, but the front desk people may not have those skills and have to translate it into Spanish. If you look at countries in Africa or South America that need intelligence and medicine, we built it for that

Douglas: 00:48:19
specific purpose that it can go into any country and suddenly bring the level of education at the point of care to a much higher level. But yes, it's extremely important.

Hanh: 00:48:32
There's a question here on Medicare. So can we take that? "Curious to know how Medicare audit process might deal with AI?" What is your take Doug?

Douglas: 00:48:43
Well, I was on the leadership of the Florida Orthopedic Society when the CERT audit came to Florida and said 75% of the total needs were not medically necessary and actually presented with the me, the Medicare area contractor, the new protocols. That's actually what pivoted us from diagnosis and treatment to medical necessity criteria because we can use that

Douglas: 00:49:01
same engine for diagnosis and treatment. We can meet medical necessity, CRI criteria for Medicare. So that's what we did first before, prior auth. My friends in plastic surgery, uh, cardiology, were all getting audits and suddenly now those go away. If you're using the system for prior auth, it's the same thing.

Douglas: 00:49:24
Now, the flip side of that, let's say you're a hospital and you don't wanna change anything. We're, we're now working with payers to use those same intelligence tools to do audits. In other words, you're not meeting medical necessity. Fix that, where there's gonna be large clawbacks.

Douglas: 00:49:42
And that's probably going to be the moving force in many hospital is using AI for audits for things that they're not doing right in the first place. Whether it's overcoding, whether it's day-to-day within the hospital, the, the, uh, work in the er, the work in the offices. So it's a two-edged sword. It can be an audit tool or it

Douglas: 00:50:01
can be a compliance tool and tell you immediately if you're meeting compliance or not meeting compliance on both sides of the defense. So yes, it definitely applies to Medicare and we pivoted a number of years ago, specifically for the local coverage determinations of Medicare.

Hanh: 00:50:20
Mm-hmm. Mm-hmm. Thank you. Thank you for your feedback there. So now, Conor, there is a question here. Maybe you can help. "I'd like to hear how McKinsey are pitching the use of AI to clients in the public sector. Are there grants and reimbursement

Hanh: 00:50:34
opportunities in place yet?" What do you think?

Conor: 00:50:39
That's a good question. I, I don't know. I know my, my wife is at McKinsey. I mean, I know McKinsey does an absolutely phenomenal job in this space. Well, I kinda have to, I mean, I don't have to say it, but my wife is in AI at McKinsey, so I'm going to say it. But I actually also know that they really are doing an amazing job, and I

Conor: 00:50:55
think bringing, uh, bringing it in, I've had a lot of conversations with some of their, uh, their top folks around all this and, and how they're doing that in terms of the actual, like inner workings and what they can, they can do. I, I don't have tons of insight around that. I just know that. When we are, um, there, there's so

Conor: 00:51:10
many different approaches to using Generative AI in different companies. And again, sort of like when I'm, when I'm going out and I have a, a YouTube channel around all this, kind of talking about the different use cases as well as, as the LinkedIn, uh, examples on my LinkedIn and everything else, what I'm finding is that the use cases are going to be so different from everybody.

Conor: 00:51:32
So I think that a lot of companies will expect Generative AI to come in and be layered on top, and then everybody uses it. And I just don't see that. I mean, I, I see how you can use it, but that feels like a very base level. It feels like you're, you know, putting in Salesforce and Salesforce is going to replace this.

Conor: 00:51:48
And so that old software is taken away and everybody just has to learn on this new model. And that, that's just how you do it. That's like a train on the tracks. This does not feel like a train on the tracks. This feels like an all-wheel vehicle where you can go wherever you want and do and explore this entire world,

Conor: 00:52:00
not just a train on the tracks. And. What I mean by that is that it's going to be different for not just every company, but every department in that company and every individual in that company, they're going to be able to use Generative AI as long as they can understand that it goes beyond just, you know, what Microsoft and Google and everything

Conor: 00:52:21
are gonna be putting into your email. I think that's all the table stakes. I think that's all the tools where, you know, in Excel you're gonna be able to write in natural language and in email it's going to, you know, auto-fill all this stuff for you. Those are all great and that's more like a tool, right? So, but that's not what

Conor: 00:52:35
we're really talking about. What we're talking about is this Generative AI and how as it exists right now, and it's even in its early days. In order to fully take advantage of it. And it's sort of like what you were talking about earlier. You have to understand how to have a conversation and make your work more productive.

Conor: 00:52:52
Only you understand what's holding you back in your work. And by the way, these are even things that you think about holding you back. You just naturally go into your work and you spend the first three hours on emails. And you know, and especially I work, I, by the way, I train, uh, in a ton of giant healthcare, some of the biggest healthcare companies in the world.

Conor: 00:53:09
I go in and I train their, uh, their senior leadership on this because healthcare is huge on this. There's so much to do, so much paperwork, so much everything else. And as Doug was sort of sinning, as soon as you start automating, uh, all of these things, it actually frees doctors up and other administrators up to do the work of interacting with people.

Conor: 00:53:27
And the stuff like, think about it, medicine is just a function of people getting hurt and coming to somebody else for help. That's what it boils down to. And when you allow doctors who are fantastic at this, and orthopedic surgeons and everybody else to automate some of those menial tasks, And to actually spend more time on the patient.

Conor: 00:53:44
That's where it really is a game changer. And that's what I would extend across every single function. Think about what you do in your work and ask ChatGPT how it can help. And I think the answers will stun you.

Hanh: 00:53:54
Mm-hmm. You're absolutely right. It's so personalized. You know, I ask myself often, like, what is my pain points? What is it that I'm trying to do? But I struggle, right? Or what is it that I hate doing? So from there I said, well, ideally, how, how do I want this to be?

Hanh: 00:54:10
So then, you know, I, I ask ChatGPT to help me ways. And again, it's a good baseline, but it all depends on how you ask, how descriptive. Again, it's in your prompt engineering. That's, that's everything. And now that there are plugins in ChatGPT, including Wolfram, which is the everything in my book, right?

Hanh: 00:54:31
It's, uh, it's a great plugin if somebody knows. Wolf frame, keep exploring and learning about that. It's a lot of science, technology and engineering. And then there's BabyAGI, AutoGPT, Hugging Face, Python, Pinecone. I mean, all of this, I know it sounds kind of foreign, but if you're

Hanh: 00:54:52
gonna be integral, it's good to have some level of understanding. Maybe you're not gonna be a developer, but it's good to understand because by understanding some of the capabilities it will uncover of what it can do for your inefficiencies. Right?

Conor: 00:55:07
I think that's the key, what you just said, because Wolfram, I have a, I have a YouTube video on using Wolfram because I didn't even know what Wolfram was. And the really cool thing is it just codes. You know, you talk to a natural language and it says, Hey, guess what, Conor? You know, it'll talk to me.

Conor: 00:55:21
Cause I'm like, Hey, I'm Conor Grennan and it'll be like, Hey Conor, great to meet you. You want me to call you Conor, Mr. Grennan? I'm like, oh, call me Conor. It's a very natural sounding thing. But the really cool thing is I went in there and used these ChatGPT plugins, code interpreter.

Conor: 00:55:33
Things like that. And I demo all this stuff in these YouTube videos. But one of the things about Wolfram I was so amazed by is you'll ask it to complete a task and it will say, okay, I'm gonna go, and it will show you the Python code that it's using to do that. And the really cool thing is that there's times where it finishes and

Conor: 00:55:49
then says, Hey Conor, that didn't work. I'm gonna try something else. Here I go. And then it'll try something else. It's talking to you as if it's a coder sitting next to you. So I encourage you, as Hanh is saying, try these things out. Because here's the thing, folks, this is natural language, meaning

Conor: 00:56:05
you do not have to know anything. You just have to go and ask ChatGPT, who's the genius of all geniuses, Hey, I'm looking for an AI tool that will help me do this, or I wanna do this, or how would I even think about this? Talk to it like you're talking, like I'm talking to you right now. And it will help you do all those things.

Hanh: 00:56:22
Mm-hmm. And I wanna share with you a perspective. It can be intimidating. Okay? So let's face it. It's still new for everybody. We're here to share our experiences, our learnings, and our triumphs and a lot of failures along the way. So it's okay when you first try and

Hanh: 00:56:37
you don't get what you are looking for. Keep trying and, and there's a psychological part of it that you have to overcome. At least I did. And once I overcome it, that instead of thinking, well, it's kind of useless cuz it's not giving me what I'm looking for. Well perhaps I need to learn better. Perhaps it's a wrong plugin

Hanh: 00:56:57
that I'm trying to dig information from and so forth. So there's a psychological obstacle that you're gonna have to challenge yourself. Okay. And it's okay if it doesn't work out. Keep trying. You'll find a solution to other means, whether it's it's Wolfram, AutoGPT, BabyAGI, all these terminologies.

Hanh: 00:57:19
You're probably wondering what that is. Well, it's good to get an understanding of that. There's a question for Doug. Hang on here. I had it right here. It keeps moving. I'm losing sight. Okay. I'm so sorry.

Hanh: 00:57:33
Every time there's a new comment, it pushes it. Okay.

Conor: 00:57:35
Well, let me just push to Doug too, sort of, you know, Doug, when you're developing all this, right, I mean, when I think about how ChatGPT works and how it reflects your experience, I mean, Doug, you're building these incredible things, right? So how has sort of something like AI sort of just, you know, augmented your own process, have

Conor: 00:57:51
you found ways of doing that? Like where, you know, you're great at this, but this is actually condensed your amount of time in, in, or workflow or helped your work workflow?

Douglas: 00:58:02
Mm-hmm. Well, it's like, uh, the issues of, uh, both prior auth and care pathways and what is the latest treatment for disease X, Y, and Z. Uh, what things should you be doing or not be doing? It's wasteful. Or prior auth. Um, I, I feel sorry for

Douglas: 00:58:18
the pain clinic guys. You read the, the, uh, policies from the different payers, somewhat. Six weeks of, uh, conservative therapy. Another one's eight weeks, another one's three months, another one 12 months. Uh, knee arthroscopy is a good example. Uh, Florida blue here has over 400 data elements in their policy. Humana only has 46, and

Douglas: 00:58:39
they're all different. And they're not in the E M R, probably never will be, but you have all of these different policies that, you know, Texas Blue doesn't even require prior authorization. But those are specific examples of trying to sort out all of the noise to make it efficient. And that's kind of what we've worked on,

Douglas: 00:58:58
not from a ChatGPT, we're using that now. But from just organizing information, distilling it down, structuring it so the output is structured at a, at a level, you'll, you'll never see in the EMRs. You can actually have predictive analytics, predictive prevent. That's where we're going with the system. But it was, we're now going back to the original diagnosis

Douglas: 00:59:25
and treatment protocols. Now that we've pretty much finished off and taught the machine all the different payer protocols as a conversation, you literally have to rewrite the protocol as a question answer system so that when you, when you finish the conversation, you know, you've met criteria, not met criteria, and it writes you note for you. So it's a, it's a different type

Douglas: 00:59:45
of system that my medical assistant use it, I don't even use it, but if I'm doing treatment than I do use those things in the emergency room. So it just depends on what you're trying to accomplish for coding, uh, a good example, we, we had a, uh, healthcare executive have a, uh, a stent put in his heart and I said, well, we can probably teach you to, um, code your

Douglas: 01:00:11
own procedure in ICD 10 PCS, which is beyond the Post-it notes in the coding department. He was able to code his own procedure with no coding experience, but just answering questions. Um, so if you look at something like CPT codes, they're analog. They just come in a number and there's a few groups.

Douglas: 01:00:30
If you look at ICD, iCD 10 PCS, procedure codes for inpatient, there's real logic behind it.

Hanh: 01:00:37
Okay. Here's another question from Arthur to you, Doug. "What is your, uh, take on the possibility that AGI replacing orthopedic surgeons and when do you think?"

Conor: 01:00:50
It's a hard question to ask Doug. I gotta say.

Douglas: 01:00:52
Well, in terms of diagnosis and treatment and teeing somebody up for surgery, I think there's gonna be a big application because if you fast forward to 2030, there are not enough orthopedic surgeons in the country to do total knee replacements. Not, not counting any trauma or, or sports medicine. Oncology, any of that.

Douglas: 01:01:16
So I think you're gonna see the surgeons in the operating room and maybe not even in the office where they're really trained, use intelligence systems and augment, uh, information where the office can really tee people up for surgery and clear them for surgery, both medically and for policy. Um, but in terms of robotics, there are certain types of robotics in orthopedics.

Douglas: 01:01:41
Um, Some of them are really good, some of them are marketing toys. It just depends.

Hanh: 01:01:46
Mm-hmm. Here's another question from Alan. "So, what do you think, Conor, how valuable is it right now for people to know when they are interacting with AI?" I see many companies trying to make chatbots feel or appear human. Would it be better to clearly discuss when healthcare advice is coming from AI or education?

Conor: 01:02:09
Yeah, that is, that is a great question. I think it really depends, uh, on whether or not, so for example, I was listening to this, uh, the great podcast. I have nothing to do with it, uh, called the Cognitive Revolution. But you can get smart really fast on it, uh, and talking, um, about these, you know, legal, uh, you know, sort

Conor: 01:02:26
of services that will automatically write, you know, letters to, you know, the judge or just sort of, you know, uh, try to dispute a parking ticket or something like that. And if there's regulations, you know, for example, in Europe, which is much more regulated. Saying, we have to know if this is AI or not, that actually

Conor: 01:02:44
hurts the consumer, right? So that hurts the consumer by saying, Hey judge, this is an AI driven thing, when in fact the consumer is really just trying to dispute something that is really well within their rights. So I think in that case, maybe it doesn't have to be, you know, discussed or disclose that this was AI, for example, like I'm on, again, I have no affiliation

Conor: 01:03:01
with these places at all, but you know, American Airlines seems to have a chatbot on their Twitter or something, or at least it sounds like ChatGPT and the responses are instantaneous and fantastic ChatGPT tends to be extremely empathetic. It reflects you, but it tends to be very kind and eem. No, it's, I'm sorry. It tends to behave very

Conor: 01:03:16
kind and empathetically. I like that. I think that's kind of a nice way to be. Now with doctors, it's kind of unbelievable, right? And Doug has probably seen the statistics as well. Patients tend to, in, in non peer review papers tend to prefer, uh, you know, the responses of.

Conor: 01:03:36
Something like a ChatGPT two doctors, they tend to think that ChatGPT may be, may be more empathetic and responsive and everything like that, right? So, so in that case, do you really care whether it's that, I mean it almost is. I'm wondering if in the future it almost sort of, look, we can think of a lot of times in the past where I wanna hear my diagnosis from somebody who looks

Conor: 01:03:57
like me, sounds like me, the same race, creed, Bo, et cetera, et cetera, right? Is this just going to be sort of another form of bias where this, you know, AI doctor is giving you the best advice that they can give you, but people don't wanna hear it? Now, maybe that's just a personal preference and I don't wanna comment on that.

Conor: 01:04:13
But what I will say is that a lot of the best information comes from AI, but people will be e little bit hesitant. So I think it really depends. So for example, if I'm getting, you know, brainstorming ideas on a market, so again, I work with a lot of marketing teams. If I'm helping marketing teams, you know, use this, you know, with prompting, as you said, uh, you know, to get great

Conor: 01:04:29
ideas, I don't think they really care. People do tend to care. If it is something around, um, your life coaching or something like that, because it feels like it doesn't really understand you. But I'll also say this quick example. My daughter's 12 years old. She dyed her hair red the other day, which is, you know, fine, but okay.

Conor: 01:04:47
Anyway, she kind of got red hair dye all over the place, which is also fine. Anyway, I sort of asked Bing, which is powered by ChatGPT four, How do you get red hair dye out of the walls? I know, I know. It told me that. But then at the end I was also like, Hey, you know, I also wanna be a good dad and I'm sort of a little

Conor: 01:05:04
frustrated, but what should I do? And, and Bing is like, great that you're trying to be a good dad. Here's some advice. You know, like your daughter's just trying to be independent. She's stretching. You know, she's, she's 12 years old, she's feeling, you know, all this kind of stuff. But just remember maybe that if you could

Conor: 01:05:15
talk to her about this and talk to her about responsibility, but listen, don't bash that creativity down your daughter. And it was giving me the best parenting advice I'd ever gotten in my life. So anyway, point being, I would not, I mean, I think some things should be labeled as AI. I saw some other things in the chat about like how you cite things.

Conor: 01:05:34
That's a different thing. You know, I think in education, I think citing is important. Very, very critically. If you do something in the chat or with AI, often you won't own that copyright. So guys, be careful out there. If you are writing something for marketing thing, if it is determined to be by AI and you've just written this

Conor: 01:05:47
giant thing, you copyright for Coke. It will not belong to you or to Coke. So be careful around copywriting. It's hard to detect, but still, I guess my point is that use AI responsibly in that way, but understand that people are going to want some of their information from AI and some of their not. But also, I don't think it should be a blanket statement of if it is

Conor: 01:06:09
done by AI, it needs to be disclosed. Cuz I don't think that's always in the best interest even of consumers.

Hanh: 01:06:13
Mm-hmm. I agree with that. And also the quality of your chat bot is contingent on what's under the hood. Right. And under the hood it's everything in the quality of your data, your data sets, how well it's a refined tune, cleaned up and so forth.

Conor: 01:06:30
You know, Doug, Doug, if I could just sort of say, Doug had a great point on this, Doug. If, if, sorry, how? I just wanted to kind of get Doug's take on this because Doug, you were talking about the criticalness of critical nature of like how clean your data is and I was wondering sort of like how you think about that.

Conor: 01:06:42
If you could kind of like enlighten us on the importance of having, you know, good data going into these systems. Cause I think a lot of people would be interested in that.

Douglas: 01:06:50
Well, we go to trusted sources, whether it's CDC, uh, I mean the machine can read the yellow book and infectious disease and give information on all of the different countries that you might be visiting and what time of the year for that specific disease you need to be, uh, concerned about. It might have the testing for certain things.

Douglas: 01:07:09
What's your immunization should be. Uh, so you get all this information, get 'em from trusted sources. That's probably my biggest worry about some of the ChatGPT. I think it's fine up to now where no one's tried to poison the water it's drinking from, but in the future, my worry is the hackers are gonna go out and book this information and

Douglas: 01:07:26
if that information is used, that's why the references are so critical. But yes, that, uh, we look at trusted sources of information, uh, whether it's N I h, whether it's CDC, that's what we really look at. We look at FDA labeling. Those are real information with real sources. Um, so that's where you

Douglas: 01:07:48
really have to start from. But even textbooks have, uh, biases. Uh, everyone has biases in some way, and you have to kind of filter that out. The subject matter expert has to review the algorithm. At least that sort of editing goes into our system.

Hanh: 01:08:05
Mm-hmm. Mm-hmm. Well, even ChatGPT, right? It's not a hundred percent. It's, it has its weaknesses. So even if you are trying to create your own chatbot, You have to recognize that. So let's take another question here. "All of the content that LLMs are learning from, its original", let

Hanh: 01:08:22
me see, lemme just write that here. "So all the content that our LLMs are learning from its original content that humans have created is an internet gets polluted by predominantly AI generated content. Is there a risk of stagnation due to lack of sufficient and original content?" What do you both think?

Douglas: 01:08:42
I think the best example is, uh, look up Silicon Valley Bank. If it was bought before it went, uh, under, uh, that's a good example. The, it was never taught what happened to the bank. And it'll give a description of, Hey, this is a great bank in Silicon Valley and you should use it and everything else and doesn't exist.

Douglas: 01:09:02
And it's current floor. It's just past floor. So I think timing is everything. The data is everything. Uh, you really have to look where that information came from. And if you notice it came from a site that was dated in 2022, then you understand that output, but you have to understand where it came from and that

Douglas: 01:09:23
the data's not contaminated, biased, or just plain and misinformation.

Conor: 01:09:27
Yeah. I, th, I think that's a great, I think that's a great point, so understanding where it's pulling its data and then, and then this, um, this question, I think it's from Prashant down there is, is a great one. It's sort of almost more of a philosophical, right? Like what is creativity?

Conor: 01:09:41
And, and the internet, by the way is, is huge. So I'm not sure we're gonna run out of that. But I've also seen, and I think that people are see in ChatGPT four, it starts to make, uh, you know, to draw insights from each other. So example, if you know, for example, if you use any, uh,

Conor: 01:09:56
text to image type of software. So I use Mid Journey, for example. And you say, you know, make a, make the Sistine Chapel as if it was balloon animals or something like that. It can, it can do that. And I think that it could generate a, it could have generated that prompt itself. If you said, give me 10 really, really creative prompts, it can draw that.

Conor: 01:10:14
So I'm not sure that creativity, and by the way, I come from a creative background as a, as a, as a writer. But, but I'm also not sure that creativity isn't just drawing unexpected, uh, connections between unparalleled things. And that, to me, if that's the source of creativity, I believe that this can do that as well as a human, I mean, just look at the idea that, you know,

Conor: 01:10:33
if you ask me to come up with 10, uh, new, you know, crazy ideas for, you know, the next Pepsi commercial or something like that, and then you ask ChatGPT to come up with a hundred, it's gonna do that much quicker than me. And I'm not sure that mine will be any better because I'm, I have a human brain, so, I'm not sure that we are going to think about.

Conor: 01:10:53
I don't, I don't know about thinking about it as, as polluting the environment. I think that there's, you know, Doug sort of referred to, there's so much bias on the internet already and so much chunk on the internet already. I'm not sure that ChatGPT can't act, um, creatively and actually expand our creatively.

Hanh: 01:11:08
Mm-hmm. I do too. Here's another comment. "Prompt Engineering is a glorified title. It is just a way to ask questions better so you get better answers." It's true, it's true. But that comes with practice, right? The more you use it and the more that perhaps you don't get the

Hanh: 01:11:25
results that you want, you're gonna become a better engineer. Prompt Engineer, because that's how you would get closer to your output. So now, here is one, how might Generative AI revolutionize healthcare education by simulating rare and complex medical scenarios, allowing students who gain hands-on experience and develop critical thinking skills

Hanh: 01:11:49
in a safe and controlled environment? So what do you think, Doug?

Douglas: 01:11:53
Well, I think if you go back and look at searches, we all had to tailor our search, uh, engines to specific questions and make sure we were looking for the same thing. If I type in fibromyalgia today on Google, it'll come up with probably, you know, 12 15 million. 12 15 million, uh, searches. If I go to ChatGPT, I'll get

Douglas: 01:12:14
a curated answer for that search with the references. In terms of education, I mean, patients and doctors go to Google all the time. Students go to Google all the time. But at the same time, if you look at that information, it needs to be curated. It's not really curated. Um, so I look at it for education. Suddenly you can get the

Douglas: 01:12:40
information you need when you need it, and be a trusted source. That's the key.

Hanh: 01:12:46
Mm-hmm. Mm-hmm. Here's one for Conor. It says that, uh, let's not forget some existing challenges. Access and equity data, privacy and security, bias and fairness, teacher training and support and ethical considerations. We need solutions and

Hanh: 01:13:01
yes, we can find them. So what do you think?

Conor: 01:13:06
I mean,, 100%. I mean, like, uh, I think in, as I'm sort of watching the chat too, you know, it's, um, The ethical considerations are critical here, right? There's gonna be a ton of bias, and as I'm just saying, like references can be fake. This is not a good knowledge tool, right? It's a good learning tool. Like anything that's sort of like

Conor: 01:13:24
almost contemporary knowledge, you can't really trust, which is why we have to ease into this. It doesn't feel like a, you know, black to white sort of thing. We sort of have to, when we are learning and when we are educating ease into it, start using these tools, start integrating them into the classroom or in your work, start integrating them.

Conor: 01:13:43
Where you have to be really, really careful is, are you, uh, you know, putting things in that, it's sort of like this case with the lawyer that just like put in all these, you know, legal citations that actually didn't exist. So you have to be so, or if you say, Hey, gimme a quote that, uh, you know, Jack Welch from, you know, GE said to his people three years ago,

Conor: 01:14:02
and it'll give you something great. What are the chances that he actually said that? I don't know. It's probably about 50 50 now. ChatGPT four hallucinates on a very, very small scale, it's probably about 4% or something like that. So, but still that's 4%. I would not go into a meeting

Conor: 01:14:14
and say, Hey, I'm gonna meet Doug Dew uh, for a coffee. Tell me everything about it. It will tell me everything about Doug will be real and accurate. Probably not, and I'm probably gonna embarrass myself. So I think with ethical considerations, you really have to understand what's real and what's not,

Conor: 01:14:27
and how you use it responsibly. And again, and I agree with some, uh, in the chat too, like it's. With AI, it is always good to, you know, reveal, you know, especially when you're using it and everything else. But I think that if you're just using it in the same way, you might use a brainstorming partner, I, that doesn't feel that critical to me.

Conor: 01:14:44
It does feel critical when you are talking about taking something wholesale and you know, putting it in as your owner or something like that. Those are ethical considerations that I actually have pretty strong opinions on. At the same time, I can't control what people do any more than I can control if they're cheating on their taxes. So that's why I don't get

Conor: 01:14:58
too bogged down in it. These are really individual considerations, but I do think that regulation is critical, so at least the consumer knows when the consumer is protected.

Hanh: 01:15:08
Mm-hmm. Mm-hmm. Very good. We're gonna take one or two last questions as we are near closing. So, for the pharmaceutical industry, what changes do you predict in the coming future of AI? What do you think, Doug?

Douglas: 01:15:22
Oh, I think you're gonna be seeing transparency that the PBMs may or may not like. I think for employers understanding their real true cost, what the rebates are for pharma, whether employees can go to the pharma and get it filled the first time and not wait two to three weeks to get the medication filled. Or say somebody being discharged with

Douglas: 01:15:44
depression from the hospital, what if they can't get their important medications? Because all of the payers have different rules even though the FDA labeling is the same. So I think you really have to look at it from a standpoint of what are you wanting to accomplish and go to that goal. And again, that's why I say any algorithm needs to have a specific mission.

Hanh: 01:16:11
Mm-hmm. Mm-hmm. Thank you.

Douglas: 01:16:12
And for pharma, it's getting the patient treated at the lowest possible cost, but appropriate care. And cuz in case of an employer, it's their money.

Hanh: 01:16:22
Mm-hmm. Mm-hmm. Thank you. Thank you. So I wanna acknowledge, uh, response from Hannah. "So it's more about how you use the answers. It's never been just about getting the answers, it's understanding them,

Hanh: 01:16:35
analyzing them, and restructuring them." So, absolutely, I echo that. So thank you. And another one, "Working with AI and using it speeds up massively of your work." This is from Pawell, "So, still needed to revise some and keep, keep it safe for patients." Absolutely.

Hanh: 01:16:52
Absolutely. And then also another comment, "In our education system, just a, com, competition between individuals, you know, is our education system just a competition between individuals using ChatGPT should be encouraged. It's not cheating if the youth has this same tool." Absolutely.

Hanh: 01:17:12
It in my mind, it is not cheating. It is. It is your tutor, your personal tutor. It can enhance or bring out the critical thinking and it can help you grow that. So I think it's great. So we're near, at the end, so I'm going to ask Doug and Conor a few more questions. So, Conor, what hurdles do you foresee in the integration of Generative AI in your

Hanh: 01:17:35
areas and how, let's say, how do you think you're gonna prepare to overcome them?

Conor: 01:17:43
Yeah, so I'll just take, uh, well it's funny, I was about to just say education, but it's really a human nature problem. So whether I'm gonna talk about education or whether you're in, uh, banking or healthcare, anything else, cuz I deal with all these, uh, companies all the time. It really is just the limitations of ourselves and

Conor: 01:18:00
I think that limitation comes. In two ways. Like number one is the validation of our entire existence and career. So I find this a lot with faculty members and professors and teachers. They've spent decades teaching something in a certain way. Learning is something in a certain way, and then they're asked

Conor: 01:18:16
to do something differently. That's a very human nature thing to rebel against that. And I really wanna be, uh, sympathetic toward that. And I think you can, you know, mirror that across anybody who's been in management tech for a long time, learning and development, marketing, anything, you know, even CFOs and everything else.

Conor: 01:18:34
They have learned something in a certain way and it's very hard to break out of that. So I think it's unreasonable to think overnight everybody's gonna change their way of doing things. Now I sort of tried, but that's because this is really my life. Now. The other thing I think that

Conor: 01:18:47
kind of runs into obstacles is literally how the brain functions. So again, you know, the brain survey, as I said, sort of like does pattern prediction and automation very, very well, right? So when we see something like Google, right, we know that there's a command response. You ask Google for something, ChatGPT is not a tool.

Conor: 01:19:05
It behaves like a human. And so when I think about the obstacles that people are gonna run into, it's our own brain. This is a lot of the training I do is saying it's not about the tool, it's about how our brain functions when we see this. So just to give a very quick example around this. If we see Google, and we've known Google

Conor: 01:19:19
for, you know, the last billion years that Google has been around, and then we look at ChatGPT, we think, well, this just may must be a better Google, right? Because everything else, like when you see the quickest adaptations of. All these new things, Spotify, Netflix, Dropbox, uh, Instagram, you know what you're upgrading from, right? You know, you're upgrading from,

Conor: 01:19:38
uh, you know, a old filing system to Dropbox or hotels to Airbnb or your CD collection to Spotify. What are we upgrading to with ChatGPT? I don't think anybody really knows. So people think, well, it must be better Google. It's not, Google's better Google. Let Google be Google this. You talk to like a human, but your

Conor: 01:19:53
brain portrays you over and over again. So when you see ChatGPT, you think, I'm gonna ask you for the top 10 places to go in Costa Rica and it'll give you something and you'll walk away, which is foolish. It's sort of like if you go to Google and ask for the top 10 things to see in Costa Rica, it'll give you a 2019 blog and you'll take that as opposed to ChatGPT, which will act like the

Conor: 01:20:10
head of the tourism board and talk to you for as long as you want. So I think our brains really betrays. And so I think that that goes across industry, which is how our brains function with this. We have to get outta the mindset of treating this like a tool and into the mindset of talking to it like a human.

Hanh: 01:20:26
So true. So what is your take, Doug? What hurdles do you foresee in the integration of Generative AI to your area, and what do you think that we can do to prepare, overcome it?

Douglas: 01:20:36
Well, I think the hurdles in medicine is no one wants to change. You talk to hospitals, you talk to doctors, you talk to payers. Uh, change is hard and it has to have real improvement to make that change. And if you don't offer real improvements, there's no reason to use the technology if it doesn't move the needle. I mean, I look, I look at it as the

Douglas: 01:20:57
CMS performance, the quality measures. Uh, I served on our academy, uh, quality measures committee, but it really wasn't moving the needle. Uh, there was a test called a SANE score, which just says, an outcomes that patient reported outcome. Well, in effect, every doctor asked that, every day, are you better, worse of the same, and it may or may not be recorded.

Douglas: 01:21:23
So it has to be something that moves the needle, not a bureaucratic step in a process so somebody can check a box. So that AI needs to be useful, it needs to be valuable, it needs to improve the workforce, not a workflow, not make it worse, and it needs to be curated so we can have it as a trusted source. And going forward, I think this conversation a year from now

Douglas: 01:21:48
will be completely different.

Hanh: 01:21:50

Douglas: 01:21:51

Hanh: 01:21:52
Yeah, things are evolving so fast that next month we'll have a different conversation of its growth, further growth.

Douglas: 01:22:01
Mm-hmm. Agree.

Hanh: 01:22:01
Okay. Same question for both of you as we wrap up. Could you share a key message to our audience about the importance of understanding and embracing Generative AI? Go ahead Doug.

Douglas: 01:22:14
Well, I'd say look at all AI applications, whether it's deep learning, machine learning, Uh, really geogen, uh, Generative AI is important and also expert systems, but understand the difference, understands what their pitfalls are under where, understand where it's going, uh, and who's the trusted source, the developing it. If you have somebody that can really

Douglas: 01:22:36
be trusted, is ethical, putting that information together in a useful manner, that's probably more important than their degrees or where they train from. Uh, that's the key to me as AI, uh, is who is leading that charge in medicine, I think you need to have a co-director of any medical AI program, otherwise you lose track of things. And what I mean by that, um,

Douglas: 01:23:01
imagine you're building an e emr. The computer programmer puts in that, Hey, I'm gonna stop antibiotics 24 hours after surgery. But what happens then if it stops the antibiotics on a septic patient, they die. And if you're not a physician, you don't understand those, you don't get down in the weeds with the details that are important, and

Douglas: 01:23:24
that's what I thinks important. And you can get those details on ChatGPT and again, have the reference material.

Hanh: 01:23:31
Thank you. Thank you. Go ahead Conor.

Conor: 01:23:33
Yeah, I would say the big takeaway for me is just, you know, try this thing out and find it useful. The one thing be careful about, be careful about entering, uh, you know, proprietary data from your company. Don't, don't kill yourself on that because you know, again, it's sort of like, you know, it's like the Samsung example. It's like pouring a Coke into a

Conor: 01:23:50
swimming pool and expecting the swimming pool that tastes like Coke. When you put in a data set, it's not like you put it in as a library and somebody else can pull it out. That's not how ChatGPT works. So I don't want people to get freaked out by that. But do be careful because of Doug knows very well, like third party agreements

Conor: 01:24:06
and HIPAA compliance and everything else. Be careful what you put in there and be careful if hallucinations in that. If you ask it for anything sort of contemporary. If you look up any one of us three and say, what do they do? Or what's their family like? It may not give you correct information, but with everything

Conor: 01:24:19
else, just test it out again. Sort of like on my, you know, LinkedIn, on my YouTube channel, everything, I just have a ton of just examples. Like, Hey, here's how you use it. Here's what you do. Just try this, try this, try this, try this. And that's why some of these things went so, you know, had millions of

Conor: 01:24:33
views and everything else because people are like, wow, well try that. Because it's that creativity. I saw this in the, in the chat as well. It's gonna, if you're more creative with it, it's gonna mirror you. It's linguistic matching. So this does that very well. It will match your who you are and what you do.

Conor: 01:24:47
So if you want to be more creative, be more creative with it, just try it out, find examples, try things out. Just test drive. That's, that's sort of how I think you get the, how you really move the needle.

Hanh: 01:24:58
Thank you. Thank you so much for both of you. And thank you for the audience. And you can follow Doug, Conor on their LinkedIn and the description of their profile is on the event as well. And thank you so much for your attention. As we conclude this insightful link LinkedIn event, let's reflect on the potential of Generative AI in

Hanh: 01:25:18
the transforming field as diverse as healthcare and education. We've listened to these two about AI's crucial role in enhancing patient care and fostering price transparency. We've also seen how AI combined with medical lasers is making tangible difference in patient outcomes. AI's potential in streamlining FDA compliance in healthcare

Hanh: 01:25:42
is also clear in education. AI is reshaping learning experiences, customizing educational content, and leading us to the cusp of major transformation. We've learned about some ideas and integrative AI further into the curriculum of NYU Stern and compelling stories, how Generative AI could potentially help missions like the

Hanh: 01:26:03
next generation Maple, reiterating the social or societal impact of AI. We're also reminded of the promising uncapped potentials of AI from creating immersive personalized learning experience experiences to providing innovative healthcare solutions. Again, let's not overlook the hurdles ahead, technological, regulatory, and ethical, that require careful

Hanh: 01:26:30
consideration and preparation. As we part ways. I like to leave with a key message. Generative AI is not just a trend. It offers unprecedented power and versatility. It, embrace it with excitement and with caution. Understand it and leverage it to the fullest.

Hanh: 01:26:53
So future topics on the LinkedIn Live is the role of Generative AI in sales and marketing. Another upcoming one is overcoming AI challenges. Navigating the psychological obstacles. Also sign up using the link in the comment below to get notified of future topics and dates and also potential opportunities to be future guests or sponsor on the show.

Hanh: 01:27:19
Follow Doug and Conor on the LinkedIn profile and thank you so much for your participation and we look forward to having you join us in the future events.