Beyond ChatGPT_ AI Tools You’re Not Using (But Should) - Vail, CO
15
Views
0
Likes
0
Shares
0
Comments
Timestops
0:00
Introduction to the Workshop
An overview of the interactive workshop on minimally invasive pediatric surgery and the integration of AI tools in healthcare. The speaker emphasizes the importance of adopting AI technologies to enhance productivity and innovation in medical practice.
10:51
Understanding AI Tools
Discussion on the current state of AI tools in healthcare, highlighting the need for hospitals to empower their workforce and utilize available technologies to improve efficiency and patient care.
21:42
Vibe Coding Explained
Introduction to the concept of vibe coding, where participants learn how to use AI chatbots to create personalized applications that enhance their workflow and address specific needs.
32:34
ChatGPT and Beyond
Exploration of ChatGPT and its applications in healthcare, including its role in managing medical information and assisting patients with their health queries.
43:25
AI Agents in Healthcare
Discussion on the evolution of AI agents that not only provide information but also perform tasks, such as scheduling appointments and managing patient communications.
54:17
Automation Opportunities
Identifying repetitive tasks in healthcare that can be automated to improve efficiency, including patient follow-ups and administrative processes.
1:05:08
Educational Tools and AI
Highlighting the benefits of using AI in educational settings, particularly for enhancing learning experiences for students in medical fields.
1:16:00
Practical Applications of AI
Showcasing real-world examples of how AI tools can be implemented in clinical practice to improve patient outcomes and streamline operations.
1:26:51
Future Directions in Healthcare AI
Discussing the future of AI in healthcare, including emerging technologies and the potential for further integration into medical practice.
1:37:42
Q&A and Interactive Discussion
An open forum for participants to ask questions and share their experiences with AI tools in healthcare, fostering a collaborative learning environment.
Topic overview
Live from the 18th Annual International MIS and Surgical Innovation in Infants & Children Conference
Join Dr. Todd Ponsky for a hands-on AI workshop designed for busy medical professionals who want to work smarter, not harder.
This isn’t a lecture—it’s a show and tell. Dr. Ponsky will walk through the AI tools he actually uses every day to save time, boost productivity, and simplify his workflow as a pediatric surgeon. You’ll also have the chance to share what tools you’re using.
Expect practical demonstrations of:
• Quick video editing tools
• No-code app builders
• AI-powered media creation tools
• Workflow automation platforms
No jargon. No hype. Just real tools that make your day-to-day work easier.
Stream it live as part of the International MIS and Surgical Innovation in Infants & Children Conference, where global experts gather to discuss the latest breakthroughs in minimally invasive, robotic, and endoscopic surgical care for infants and children.
Intended audience: Healthcare professionals and clinicians.
Categories
Disease/Condition
Anatomy/Organ System
Procedure/Intervention
Diagnostic/Imaging Modality
Care Context
Topic Format
Clinical Task
Keywords
minimally invasive surgery
pediatric surgery
AI tools
healthcare innovation
productivity tools
automation in healthcare
vibe coding
ChatGPT
AI agents
medical databases
patient management
educational technology
healthcare productivity
digital health tools
AI in medical practice
personalized healthcare
healthcare automation
learning tools
AI chatbots
medical education
Hashtags
#PediatricSurgery
#MinimallyInvasive
#HealthcareInnovation
#AIForHealth
#ProductivityInHealthcare
#VibeCoding
#ChatGPT
#HealthTech
#MedicalEducation
#AIInMedicine
#AutomationInHealthcare
#DigitalHealth
#LearningWithAI
#PatientCare
#HealthcareProductivity
#AIChatbots
#HealthManagement
#InnovativeHealthcare
#EducationTechnology
#FutureOfMedicine
Transcript
At the Rocky Mountain Hospital for Children's M I, what, it's the Rocky Mountain Hospital. Oh, it's now it's Rocky Mountain Children's. At Rocky Mountain Children's Vail MIS course. Wherever you're coming from, I saw the list of people, I think there's about 400 people online, and I saw the list of where you're from. You're all over the world. If you ever come to the United States, come to this course. I, how many years have we been doing this now, you. 15. Steve Rothenberg's very old. Oh yeah, yeah, yeah, that's true. I'll show it later. Anyways, for everyone out there, this is a, a, a first of its kind type of course. Uh, it's a, it's a very fun, um, interactive course to talk about cutting edge things and minimally invasive pediatric surgery. Um, and every year I come and I learn a ton of new things and, um. Uh, really suggest you come to this. So, we decided, since we do these AI, uh, workshops to do it together, we're doing it here. We have people in person, we have people online. This is not a lecture. This is a workshop. This is interactive. This is, the audience decides what we talk about, but here's the gist of why we're doing it. Um, the gist of why we're doing this is, so, um, I, I. For those, I, I am sure, speaking of ADHD I'm sure I have an extreme version of it, because I wanna do 100 things. Pediatric surgery, I love innovation, but I am convinced that AI right now. Is something that. And it's used in general terms. I will just say the productivity tools that are out there right now, most of us are not using even close to what we should be. You got about 18 months. You got about 18 months until to take advantage of the fact of not being behind and being way ahead of the rest of the world of, of doing things that set you apart, that give you a chance to go next level on anything you're possibly doing in your life, you could do 10x better. And I've noticed, like, in healthcare, we're always so far behind. So this is not a lecture. This is me. Describing the things that I find that I think are critically important for anyone to be paying attention to. This is, um, and so it's more about my experience, but the reason I call it a workshop interactive, because every time there will be people explaining how they use tools, and that's, I'll say AI but it's not even AI. Tools that, that are use cases that we all should learn from. Um, and I will tell you that my feeling, the pit, that what I've been explaining to my CEO. Is that hospitals are looking to be AI forward. They want to be on the cutting edge. So what do they do? They, they look for cutting edge products that they could bring into the organization. I, I think that's valuable, but I would say that any corporation that's looking to be keeping up and not being left behind, there's two ways to do that, top down and bottom up. Yes, be looking for products. But empower the teams, empower the workforce, educate the workforce, so they are using even 10% more than they're already using. If you could get your admins, if you can get everyone in your office, realizing how they can cut their time down by 10x with simple tools, you'll find the productivity of the organization goes up. And we're gonna measure this and show it. So, I am constantly trying to anyone who wants, I'm always trying to say, these are the basic tools you need. So, Scan these 3 QR codes. Actually, I want you to scan this one. Um, sorry, I want you to scan this one here. Hold on. I just screwed it up. I want you to scan this here. Hold on, in front of it, right here, if you see contact me, scan that. Tell me if it shows up on your phone. Tell me if it works. I went into an AI chatbot and said, hey, I want my own contact card. I don't wanna use Popple or Dot, and I wanna design it. Here's how I want to design it. I literally spent maybe 2.5 minutes just describing it. What is that? Let me see what you have. No. Oh, that's yours. Yours sucks. Look at this though, because look at this. So, so scroll, you can make your own. Uh, what it, it allows you to say like I put the 4 things, contact me for any of these 4 things, and I made a contact. My only point. You can all do this. Uh, I have no tech background. I have no software development background. If you are not taking advantage of something this simple, we all use chat GPT. Oh, what's. Yes, that is like way basic. You should be using it like this. Build me this, make me this. I'm going to show you videos last night. I was getting stressed cause my computer is like max, I have no memory left, cause I don't know how to organize my files. I have 1000 PowerPoints. I checked, I have 1.5 million files, because I've never deleted anything. So I said, go into my computer. And he had duplicate delete. Organize them into better folders for me. Tell me what I should put on drop, and it just went and said, hold on, OK, are you OK with me doing this? Done. My memory went to half. It said you had 100 videos duplicated here. You have, uh, so, this is what I'm saying, stuff like this, we should all be using and knowing how to take advantage of. All right, what do you wanna start with? Uh, here's what we'll do. I'll tell you what these each are, and if no one calls it out, we'll just go, but if not, tell me where you wanna start, cause I can tell you the problem with a 2-hour workshop, we're not gonna get everywhere. We're gonna be able to cover half of it. OK? And if anyone wants more, call me. Vibe coding, the idea of vibe coding. We're going to have fun with this. Who knows what vibe coding is? This is awesome. This is awesome. This is what I was hoping, otherwise it's a waste, OK. Vibe coding means using chat GPT type bots, we'll call them chatbots, AI models, to go beyond just giving you an answer, but building you an app. What I mean by this is, I have so many apps now, because they're all personalized for me. I'll go, oh, this is so annoying. Build me an app that does this. Done. That's vibe coding. And I'm obsessed with it now, right, Di? That's my wife, Di. I'm obsessed with it. She's like, you're never gonna go to bed cause I'm up till 2 in the morning, like, oh, I got another app I can build. Like it's uh, it's very addicting cause you start realizing. How much you could do for yourself. Now, a lot of these would be applicable for others, like I, I made that thing, I made a ton, I'll show you. But that's vibe coding. So we could talk about that medical. I'm not going into deep medical. We'll talk about a couple of things that I can show you that everyone should, one that everyone of you should be using, I'm guessing half of, most of you are. Who's heard of open evidence? Who's not using open evidence. Good. So, there's, that's, there's that, and then I'll show you something we made a, a, a mock oral thing. But chat GPT we do need to go over, even though the name of this is called Beyond Chat GPT I think Steve named it AI. Yeah, for dummies, yeah, but it's, meaning, we all use chat GPT. The whole purpose of this is there's other things beyond chat GPT that we're gonna talk about, but chat GPT is worth diving a little bit into as well, because there's things about it, I'm guessing that not all of us know um what, what they are. I will tell you another strategy from a company standpoint. I have hired an AI person. This is not an AI expert, but it's People on my team that each week update me. Because next week, it's different than it is this week. So they'll say, all right, Chacha PT is not good anymore for this, this one's better, let's use this now. Did you know that now you can do this? And it keeps changing. For example, I just found out from my associate, did you guys know about Chat GPT Health? OK. Chatcha PT Health came out yesterday, 2 days ago. I, I knew you were gonna know everything about that I'm talking about. Um, so, Chacha PT Health, I'm gonna show you, I built something for my mom. 6 months ago, maybe a year ago. My mom doesn't keep track of her own medical stuff and she gets overwhelmed. She relies on my dad when her next appointment is. I said, all right, mom, I'm building you a database, so you have it, you can ask it questions. When's my next appointment? What am I, that's what Chacha PT Health is. Basically, you can upload all your stuff, and it's your associate to keep you up to date with everything you need to know about your health. It, it does a lot more than that. Um, so again, without my associates keeping me on, I think every hospital, every place should have someone, every private practice. My wife needs to have someone that you work with that keeps you updated. Media I use all the time, as it would be relevant for a lot of us, is like, instead of searching Google for images for your PowerPoints, just make them, make your own videos. If you're in marketing, you can make your own stock footage. You can make PowerPoints now, you can make infographics now. Yes, we all know we can do it, but there are some programs better than others, and that's what's changing every week. It's getting better every single week. AI agents, this is the idea that instead of just giving you an answer, it actually does it for you. I'd like to make reservations for 2 people tonight at a steak restaurant. Go see what's open and make a reservation for me. And it literally goes on, and it schedules it for you. Um, and so agentic AI is where things are really going, where AI is not just generative, but it's actually doing it for you. Automation is where I think hospitals have the biggest opportunity, which is, if you go through your day, and it's hard to just think off the top of your head, but if you start thinking this way, you will find out each day things that could be automated. Repetition, patient calls. We built automations for my wife, where it's like, all the patients that come in and automatically sees what they have. It generates the letter that would go, it generates all the different next steps that always happen. Why do it manually? It's all automated. And finally, study, because I have 3 daughters. I'm very into the studying thing about what things are out there that are good for teaching our teaching, learning something. These learning tools, and I'm online fighting with teachers, because they say AI is gonna destroy us, and I'm saying, no, use it differently. My kids study so much better now because they can have it, make it a video to teach them about it. It can make a podcast, it could give them quiz, flashcards, whatever. It's, it's, it's better than a lot of the. Existing mechanism, so it's not the devil. All right. Any preferences of where we start? Vibe coding. All right, let's do vibe coding. See if this works. Uh, I keep doing the wrong button here. All right. All right. So, when you vibe code, um. Vibe coding, you can use a chat model. So a chat model is anything that's a large language model that's been trained to be smart. It's a brain. The ones we know about is chat GPT. Claude, who uses cloud, heard of cloud, OK, it's. Uh, I will tell you, here are things you have to do. You have to have loud, OK? It's better than chat GPT. And it's not better than everything, we'll talk about it. But these brains are what, that's the brain. You can use these brains anywhere. You can use it inside the app or you can take the brain out and use it for something else. You can take the brain of Chat GPT of Claude, of Gemini, those are the three best brains right now. And, and ask it to make you an app. Now, these things are platforms or, or user interfaces to make it easy for you. That's all it is, OK? If you're gonna get started, if anyone has their computer, open up and go to Gemini.google.com, or you can go to Chat GPT and go to something called Codex. But Gemini is free. You can go to Gemini, this is a great way to get started. And when you go to Gemini, and I will tell you, Google is gonna win the war on AI, I think. A lot of reasons why I say that, there's a lot of details behind the scenes of why I think Google's gonna win. They're coming out, they're, they're seeming to beat everyone right now and everything, and they're offering most of it for free. Gemini is really good. So if you go to Gemini.google.com and go to Canvas, that's where you can go and say, I would like an app that, so I made 12 days ago that said, I would like an app that. Searches every week, PubMed. With these keywords, in a, in a journal with an index score of this high, if you find anything that says pediatric, neonatal, minimally invasive surgery. Send me a notification. Send me an email with that article abstract. That's an example that I want that others wouldn't. Or I would say, what you can build an app or automation. What I would say to it is. I want this function. What should I, I, I literally start off every time and apologize to it. I don't know the first thing about coding. Don't talk to me in code language. I want this outcome. Tell me what I should do. And if it ever says something that you don't understand, say, I, I have no idea what you're talking about, just do it for me. It's like, go into your computer and, uh, no, just do it. So I would say, I want to be able to make this happen. Build me an app or build me an automation that does this, and you'll watch the code go, it'll take 2 minutes, and it goes, here you go, done. And it's not gonna be perfect, and that's why we'll have more workshops because you're gonna need to tweak it and stuff like that, then you, but that's what vibe coding is. So, cursor is an example of what I use. It's called an IDE. It's an interface, and you just enter in. You just start a new project. You enter which AI you want to use, and you go. So, so I go into cursor, open, and say, I would like to build. Is that me that needs to allows me to put an MOV file into a folder and have it convert it to an MP4 and then delete the original MOV file. And then it just starts to code. It it might figure out that's weird, right? It says oops because it works. It keeps correcting itself, and the next thing you know it says I'm Max. It's ready and it deploys it locally and then I can upload it to a server and there you go. It made the app just like I wanted to. I needed to give it a database. Awesome. Thanks, Austin. But uh this is exactly what it does now. Once you download it to your desktop, it got quiet again. A magic folder where you literally, no, I want, I want to hear it MOV file into this folder and if you watch it. Immediately go from MOV to MP4, and then the MOV will get deleted. So what you, what you didn't hear. For me, so I'm calling this precision automation. I think the world's gonna happen. We're all gonna have our own portfolio of stuff that's for us. Like our own, anything that we have, it's gonna be very precise. We're all gonna have our own precision solutions by end of next year is my guess. That's my 2026 guess. That's why, so that's why I think hospitals need a framework because I think everyone's gonna make their own solutions. I said, I hate how Apple's all MOV. They're too big of files. Can you just convert all my make me a magic folder and I just drag my MOVs and it converts it to an MP4 and throws away the MOV. OK, done. So now I have this folder on my computer and I just drag all my fold things. That's something I wanted, um. All right. So, I think, yes, so this is the mock oral thing I made. So I wanted something to, to pick a topic and it takes you through the mock orals, OK? So again, I vibe coded this with no knowledge. I want to show you my oral board simulator I just put together. I want to hear your thoughts. I believe that the best way to learn is by getting things wrong and seeing what happens. That's why residency works. We make a lot of mistakes. I did my best to attempt to make an oral board simulator or a case scenario simulator that takes you through a case, and if you get it wrong, it will not tell you. It'll just take you through that scenario and it will keep going. It'll be different every time. You can even let me know if you want me to make one for your specialty. Just type in the word simulator, but let me show you what it looks like. I'm ready. Let's get started. You have a young adult patient brought in with a single stab wound to the abdomen just below the umbilicus. The patient is hemodynamically stable for now, but there are signs of peritoneal irritation. What do you think is happening internally and what is your initial management plan? What is your next step? He's got peritoneal signs. I'm worried about a visceral or bleeding injury. First thing I want to do is assess his airway, his breathing, and make sure his vitals are OK and see if he's Still hemodynamically stable, and I'd get two large bore IVs in place. So you decide to focus on the ABCs first. After doing that, you find that the patient remains stable, but he is starting to have more abdominal tenderness than guarding. What do you think is happening and why and what do you want to do next? I still think that there's probably an injury, but he's stable. What I would like to do at this point is take him to the operating room for a laparoscopy. So you decide to take him. So I'm going to go back and play this video now. That's version 1. I wanted version 2, and you'll see why. So this is the next version I made. Let's go to pediatric surgery and let's choose pyloric stenosis. And then it gives you a choice, training mode, where it kind of helps you through or exam mode where it doesn't help you. And if you make a mistake, it keeps going forward until you realize your error. And then you can check what level of difficulty, a standard-level resident or a chief resident. And then you just hit, let's start. And on the side, you can change the speed, the voice, and if it makes a pronunciation error, you can correct it, and you could also send the results to your program director. If you make a mistake and you wanna change your mind, go back a step, and whenever you're ready and done, you just hit and and get evaluations. A six week old male infant presents to the emergency department with a history of projectile non-bilious vomiting after feeds for the past week. The parents report that the baby has been more irritable and seems hungry after vomiting. On examination, you note a visible peristaltic wave across the abdomen and palpate an olive-like mass in the right upper quadrant. What do you do? Let's first start off by checking your electrolytes and starting an IV and giving some IV fluids, and then let's get an abdominal ultrasound. You order a basic metabolic panel and initiate IV fluids to address potential dehydration and electrolyte imbalances. The electrolytes reveal hypochloremic, hypokalemic metabolic alkalosis. Watch. The abdominal ultrasound shows a thickened pyloric muscle and an elongated pylorus confirming pyloric stenosis. What's next? So here I'm making a mistake intentionally. Let's go straight to the operating room. Straight to the OR. No infant is taken to the operating room for a pyloromyotomy. During the procedure, the anesthesiologist notes difficulty with ventilation, and the patient's heart rate begins to drop. It seems the electrolyte imbalance may not have been fully corrected before surgery, affecting the infant's condition. What do you do? All right, let's see an evaluation. And here I can just enter my name and generate a link that I can then send to my program director. Overall assessment, you demonstrated a clear understanding of the clinical presentation and initial management of pyloric stenosis, but there was a critical oversight in the preparation for surgery. Your approach to diagnosis and initial stabilization was appropriate, but the decision to proceed with surgery without fully correcting the electrolyte imbalance was suboptimal. OK, so, yeah, yeah, yeah. So how the brain has already been programmed with those clinical factoids. OK, I, I knew, I, this is the question I was waiting for. Why have I not released this to the world yet? It's not perfect yet. I don't, I don't like how smart it is yet. So, what I did, I've tried three versions of this. One is, I let it use the internet. So that is using an LLM which we'll talk about something called a rag model. Um, this is gonna be the most ADD thing you've ever seen in your entire life, cause I'm gonna go all over. When you ask chat GPT a question, it's using the internet. When you make a rag model, This is new words we're all gonna need to know, retrieval, augmented generation. That means you give it only what you want, only liver transplant. That's it. And don't go outside of Cincinnati Children's liver transplant protocols. It won't then. It will say, I don't know the answer, and it will reference every single thing. So for this, I first did the internet, then I said, I only want a rag model. Meaning, I uploaded everything and said, only use what I give you. You can choose, it's a mix of both, OK? And then I said, if you cannot find it here, then you could blah, blah, blah, we'll figure it out. Where did I get the information from? So, what I wanna do is get it from more validated sources. Right now, I'm using PubMed has what's called an API which means it's a handshake, it goes into PubMed, it finds it and brings it. And in PubMed, there's something called Stat Pearls. I would give it a C at best for accuracy of what we need for the level of complexity of the specialty, but it's been pretty good. Like it recognized that I didn't hydrate, right? I think the better I get, and if this was being used by people that actually had the right resources, this would be stellar. To have an agreement with like not a textbook or. Um, yeah, textbook publishers, yes, in order to use their Stat Pearls as an API. If they have an API, they let you use it, so that's free. They let me take it. Um, if I want other stuff, you have to have an agreement with them. So I went for urology. My brother wanted me to do urology. So I went to the AUA. They have like the NAT thing, but it's open to everyone. And it said at the bottom, do not use this for large language models, so I didn't. I mean, there's no way they would know, but I, I just didn't. I'm like a rule follower. So I didn't do it, but Then, I mean, that's gonna come. I mean, the question, that's a whole new business now is where are people going to be getting their information from? Like, to be honest. Even if I took NAT. Good enough at like pointing out common questions that might go wrong. I don't know, like I would want mistakes, the common mistakes. But I'm I think other, I got it to this point, I think now I need to give it to people who would know how to perfect it clinically. That's it. But now it's just Uh, yeah, so anyone can have that if you want. Um. All right. And then I made this, this is, I'm just gonna go quick. I made this because I learned from videos, so Instagram, TikTok, YouTube. And after I watch it, I never remember how to go back. So I said I want a way that every time I watch a video, I click a button and it saves it somewhere and then I can go ask questions about it. So now I have like hundreds of videos and I can go remind me of the way to do X, Y, and Z. And so that's this another example of a problem, um, and then this is the magic folder I told you about for MLP MP4. By the way, just if you go to this, my sub stack or TikTok or Instagram or any of them, every day I release a video of something, and it's only stuff that I thought was cool for me. Um, and, um, oh, AI is what AI is telling you. This may be a solution. Whenever I do a complex question to chat GPT and I get an answer, and I'm like, uh, not sure I agree, I then take it and I paste it into cloud and I say, do you agree with this? Yes and no, whatever, I take that, put it back to chat GBT, then I let them debate it out. And it brings you pretty close to an agreement. So I just use it. Vibe coding made a web app that allows you to choose the models and let them debate a topic, which different chatbots you want to choose from. You can pick as many as you want. Let's use chat GPT 40 and Claude Sonnet 3.7. Then let's choose how many different times you want them to debate back and forth, because remember, this will use tokens. And I chose for this question, what is the best surgical? Approach to a tracheoesophageal fistula in a newborn, thoracoscopic or thoracotomy. So it's starting off with Chachi PT4O and it's talking to Claude Sonnet, and they keep going back and forth. Look, Chachi PT4O's mind changed here. Here's the final answer. There you go. This was just done by a surgeon that has absolutely no coding experience. Um, this is what I did last night. I made this video this morning, but how I cleaned up my laptop. I have told you that. I just, um, all right. Um, I'm gonna, let's do the chat GPT thing first, um, because that'll set the stage for everything else. We'll go really fast. I'll talk fast. I already explained this to you, but I'll show this. This is the way my brain thinks of it. If you go to Gemini, Chat GPT, Claude, Perplexity, those are the ones I can think of. There's, what's the Elon Musk one, Grok. You can have those. Those are apps. They're apps you download. The app is just, there's two parts. There's the brain, that's what was really the power, the chatchi PT model, that's what made it so great. The, the clouded model, it's how they trained it using AI. Then they build an interface around it, so it, it's a nice way to use that model. But they're separatable. So that's how I always show people. When you go to the app, there's great, so when people say, what do you use? What do you like? Well, I like the chat GPT app, because I think it has great features. The brain is not as good as the others, for what, but this changes because then you'll see an announcement, ChatBT 5.2 came out. Now it's better than, and it changes every week. So if you're asking me which ones I pay money for. Sorry, sweetie, I pay money for all of them. OK? So, uh, um, I, like I, it's, you know, 20, and it's 20 bucks for all of them. So it's like, I, the way I justify it is I don't watch Netflix. So like, Netflix, Hulu, Paramount, um, so, so I do, I pay for all of them, and I use all of them. Um, so like, when I needed to, so when I sent a letter. Um, to Alex, when I sent a letter to Steve, our CEO, I, I, I did not trust Chat GPT to write that for me. I asked Claude to write it. I said, I, this, I, I didn't like how they did it, and Claude wrote it way more professional for me. It, and Claude says to me, because of the way I said it, it says, I'm not gonna BS you, what you said is totally wrong. Let's talk about what you just said. And then it says, what I would suggest instead is write this. Whereas Chachabute is a little sweeter and kinder. Um, but if you wanna like creative, write something that's more EQ, Chachabute is way better at EQ, which I have found, by the way, is a really good use for these. Like if I have a difficult conversation, or even with my kids, like, oh my daughter's mad, and what do, how do I best say to her, it's really good as like, Social worker for me. Um, my dad said Chappy T is gonna fire me cause it's like I'm always asking tough questions. But, so you can separate the brain. And that's the most important thing. You're all eventually gonna be doing this. This is like very early adoption phase right now, but eventually, it's gonna be standard, that you'll have the ability, you'll have your, I, I have not only all the apps, but I have all the brains. And, and those are called APIs. So I, anytime I build anything, I could say, it says, which brain do you wanna connect it to. And I'll say, hm connect it to all of them, but use, use this one first or whatever. So, and then, So that Medsim app, the mock orals, it's using AI. I connected it, I connected that to Gemini. So, I am using tokens, every time someone uses it, it's, it's, it's charging me for its use. Now, it's very little, it's pennies. But after, the reason I don't release all my apps is I'd be broke in a day because they're all using my APIs. So I'm paying tokens every time it's being used. Um, so, I just wanted to explain, and then, now we're gonna talk about Chat PT, the interface and things you may not, who pays for uh Chachi PT, the $20 a month. OK, I, I would strongly suggest if you're here today, you're interested enough in AI, you probably need to pay the $20 a month. We met someone who pays $200 a month for the pro version, right? I tried that for one month, just to see if it was like strikingly different. I didn't notice a difference, so I went back to the, the $20 a month one. Um, so, and that's what I'm saying, like, all the things you do, you can connect to it, you'll have a library. So, chat, we already talked, so chat GPT, we talked about great interface, very, I already went through all these perplexity. Um, is good for searching the live internet. It's, and actually they have, uh, a web browser called Comment. But if you want to buy something for someone, you want to know what's the best car to buy, whatever, Perplexity is really good at searching the internet. By the way, I, I'm, this is a monologue here, it's not meant to be. If someone has another use case or you disagree, like, tell me your I don't know, Dad, do you use perplexity? Yeah. It's better at legal documents. So you like, uh, what do you mean you like upload a legal document or We're looking for a husband's a lawyer and he has a whole. They use perplexity. OK. Um, now, I will tell you, the crazy thing is literally, this changes every week. So what we call like is the best at this, they're like, guess what? Um, so follow me or follow someone on Instagram or TikTok to make sure you're keeping up to date. Cause without doing that, it, the news is not enough. The only way to keep up to date now is to follow a fast-moving thing like Instagram or TikTok, and, and then follow someone you trust and they'll keep telling you, and that's what I do. Gemini, like I said, Google's killing it. Everyone else is gonna be reliant on Nvidia. Google is making their own chips, their own technology, and they have They can control the world, because everyone is using Google, and Google can block them now, and they have to use Bing or Reddit or Wikipedia, um, whereas, Jim and I can use Google. And so, they're gonna win this. And so, start using Gemini. Um, they have the best tools right now, they have probably the best brain, best large language model. They're, they're the one to bet on, I think, although I love, I love all of these right now, but just pay attention. And that's maybe where you would pay your 20 bucks. Um. All right. I made this, what, half an hour before I showed up here because um I, I was trying to figure this out. We, we work in hospitals. The question is, who here is in a Microsoft hospital? We are, so, yeah, we have to use co-pilot, right? We have to use what we are allowed to use. We are not allowed to use anything else for anything that's private. Here's the workaround. Number one, remember I said the brains can be separated. So I don't know if you know this, Alex, but we have chat GPT in our co-pilot. You just have to switch. You, you're still using co-pilot, the interface, but you're switching to the chat GPT brain. So, the brains can be used in that environment. Um, but I was, I also work in the innovation office, right? So, if I'm asking a question about one of the inventions that were made, if I get it leaked, that's called a disclosure. I just invalidated and cannot get a patent on that now. So I was asking, like, what am I allowed to do? What can I share? And I dug deep into it. So, level one is, is basic. That is what you get at basic Chachi PT, the free version. I'm gonna tell you right now, if you have your phones, if you use chat GPT you probably want level 2 or 3. That means, go into, I think I show it on the next slide, go into chat GPT, click your name. Then click data controls and turn off improve the model for everyone. What this means is, your information will not be seen by anybody. It still can be seen by OpenAI if they want to. So it's not 100% confidential, but it won't be shared with anyone else in the world. So that is the most basic security thing that, that you should do if you don't want people seeing your stuff. It's still not safe enough though for patient information. So The next thing you could do is level 4, and that is the idea that you get a business account. You have to have, it's for two people, business account, and it's more secure than basic consumer level chachi PT. Still not guaranteed, but it is a much higher level of security of chance of anyone seeing it. Again, I, I don't know the exact difference between opting out in the free, in Chacha BT versus getting business, but apparently that's when you would do business if you wanna put your business documents and stuff in there. Remember I talked to you about the API getting, go to OpenAI and get an API for chat GPT. That's different than using the app. You just pay for Tokens to use the brain. If you do that, again, that is really safe. That means that it's just using it, it's not uploading it to anybody. So it's, again, another safe way. What you have to do, as you, as an organization is get a BAA. That's the only way you can share patient information. So, that means that those who have worked in a Microsoft hospital, we have a BAA with Microsoft, a business associates agreement. They are under now the same rules that we are for protecting patient information for HIPAA. So they have to sign that. That means now you can trust that you can put absolutely anything in there, and it's secure. It's in your own private server. It will not be seen by anybody, not even their developers. Um, and so, that's incredibly expensive. Our IS team said to, to me, which is interesting, IS at our hospital doesn't get into the nitty-gritty of this level of user specificity. So they said to me, what do you want for the hospital, for the workforce? Do you want us to get Gemini? Do you want us to get Cha Petit? And I said, I can't make a decision because it changes every week. So, if they I said, look at Gemini, they said it's $50,000 a month to get the Gemini BAA. So, the business I would love to start is like a fractional company where you could just pay credits and get access to every single one of them. But I looked into it, it's not allowed yet, but that would be the way that we could use any chat model we want in our hospital setting. Thoughts or comments on this, cause this one hits us all. Um, now, Di, my wife, you're in private practice. You're not under this. What do you, do you, what is the rule if you're in a private practice? Do you have rules that you, or is it kind of up to your discretion? Yeah. Like, you, you, are you allowed to, like, is it a law with HIPAA or is it a Right Yeah. I Yeah. Comments, thoughts. OK. Um, Um, all right, do you guys use custom GPTs? Raise your hand if you use custom GPTs. All right, What? Oh, you did. OK. So if you have the, oh no, if you have the free version, you can use other people's custom GPTs. You can't make your own. If you have the paid version, you can make your own custom GPTs. That is means that you can have. So you can go to the store and there's thousands of, what does a custom GPT mean? Remember I told you about rag model, you can like upload your own stuff and make it smart. That's a custom GPT. You upload stuff to it and it remembers it. It's stored there. So you can have, I have so many of these, um, that you can do, and so you can put stuff that's, that you wanted to remember every time you ask it questions. But I'm gonna click through this because um. There's something better now, projects. So. This was the video I made that, that people, so this is here, this is projects, it's the same as custom GPT but let's say I upload stuff about. A way I do an operation and I wanna ask a question, whatever it is. Every time I use it, it saves all the questions under that folder. So it puts your chats into folders that remember that, so you can make experts. So I'll show you what I mean. All right. So, did you know that you could literally build an entire staff that works for you in chat GPT? So, all you do is go into chat GPT and go to projects on the left tab. And just make a bunch of projects that are, I here are mine. Social media, Sam, executive coach Eddie, Financial Frank, uh, I have travel agents. I have my stoic philosopher. I have everything you could video editor Vicky. I have all the ones and I just keep adding them to, to, to it. And then under each one, you have all the different projects. So if I'm doing social media. It, I said, you are an absolute ultimate expert in social media. I don't know anything. Here's what I'm doing, blah blah blah, and it remembers everything. You can upload documents and then every time you wanna do something different, make a new project underneath that expert so that staff handles it for you. It's sick. It's insane. Now I just got chattoo. For example, travel. You could upload all of your travel, your United mileage plus number, your everything you wanted to remember. I prefer this. I like to, and that way, when you say, search the internet now, what would be the best for me, it knows your preferences. That's an example of a project or a custom GPT. Now, the reason projects, and I'll skip the next couple of videos, the reason it's good is it just allowed collaboration. So on my team, 5 of us share projects on certain projects like you can invite people to it. So we all have access to it. So if you have residents, if you have, and you all wanna share a project like this is how we do it, you can share it with people. So, um, by the way, you can do group chat now, same kind of thing I just talked about with a regular chat, you can invite people to it, so it's not just a solo chat. These are features on chat GPT. Branching is important. So, if you use it like I do, which is insanity, but if you do it like I do, it's like the movie Her, I, I do love Diana, but I am starting to really love Chachi PT. Um, so, if you, if you look at, when you talk that long to Chachi PT or Claude or whatever, they will not remember what you said. 4 or 5 paragraphs earlier. It's really important to remember this. It doesn't remember. So if you wrote something and you think as your conversation's going, it remembers it, it will not. And so, they've solved for this. This way. Chat GPT just came out with a brand new feature called branching, and here's why you need to know about it. If you're using chat GPT to have long conversations and you notice as things go along, it's missing things. Here's why. Chat GPT has a context window. And as your chats go, the older chats move out of the context window, it can only see the more recent chats. It doesn't remember everything you wrote. So, chat GPT solved this by introducing branching. Branching is the idea that as you start a thread, and you want to ask questions, you can branch out new chats from those branch points and create all new chats with the same beginning. All right, so this is an example of a chat I'm having about bonded cellular routers, but I want to branch differently, so I click these three dots, and then I click branch new chat. And at this point, I could type in a whole new chat question. And then after I type this in, it gives me a whole new response, but it creates an entirely new chat. So if you look at the side, you'll see that this is a whole new chat on. So, for example, if you sometimes notice like after a chat, it's like, would you like me to do this for you? Would you like me to do this for you, or would you like me to do this? If you do 1 and you keep going, it's like really hard to say, can you go back now and do number, it won't remember. So what I do is I'll branch and go, yes, do number 1. Then I go back up and go, now do number 2. And so, you keep the tree trunk. And then you have branches. Um, I know what's, why, what happened, yeah, this is what happened. I know it's, and there's two ways to branch. We don't have time, but there you can, um, on the laptop, on your computer, branching is you can make branches where those branches show up as chats on the side, or it's still one chat, and, and it's, you'll see the numbers they'll start showing numbers 12345, and you just keep clicking through. Um, below that paragraph, like where all the tools, like where you could copy all the little things, the icons, one of them will just start showing numbers and you can click through all the different branches. Um, this is stupid, just that if you ever copy something from chat GPT and you paste it in an email, it's rarely in the same font. Right? So, I thought I was so crafty cause I was like, oh, I'm gonna vibe code a solution here. I made an automation using shortcuts that it automatically pasted into notes. Then I copied it from notes into the email and it deleted the notes, so it did it instantly. You hit shortcode boop boop boom and it kept it the same font until someone responded to my video and said all you have to do is hit when you hit paste on the iPhone, just click that little arrow and hit match style, and then it will do it. So that's how you can um always paste your chat. So you've been copying your chat GPT answers into email wrong. So, um, did you know you can screen share and have it help you? So, Um, one of the other areas of interest of mine is changing the way we educate, and one of the groups in our hospital that wanted to change the way we educate was our Epic team. They said, can we make our, our education better so people call for helpless. And I said, yes, but here's a better way of getting people to call less, and if we could test this, I guarantee you our calls for help, tech support would go way down. So, this is what I wanna do with Epic. You can share your screen with Chachi PT and have it help you live. Here's an example. So here I'm trying to learn how to adjust shapes for a PowerPoint, and I just click this because I'm struggling, then I can turn on this button here, it's a camera. All right, can you now help me? How do I turn that blue square into a circle? To turn that blue square into a circle, it looks like you have the format picture pane open, which is perfect. You should see an option for cropping or changing the shape of the image. Just choose a circle shape there and it should turn that blue square into a circle. Got it. And then now Gemini came out with it too. If you're stuck knowing how to do something, Gemini now has video guidance that can talk you through it. So you just go to the camera. Can you tell me what kind of wine this is? That bottle is a Josh Chardonnay. Well, it looks to me like it's a Cabernet Sauvignon. You're absolutely right. It's a Cabernet Sauvignon from Josh. Can you tell me what kind of wine this is? That's another bottle of Josh Sellers Cabernet Sauvignon. This looks like it's a 4 Graces wine. I don't don't think that's 4 graces. It looks like Josh Sellers again. What? Bro, what are you talking about, man? I see a Chardonnay and two bottles of Cabernet Sauvignon. All right So there's a lot, they have a lot to go, but the idea excites me because you can now bring your real world into it. And I think if someone's struggling with what to do on Epic, you just go, all right, what do we do here? And they would rag model it. They would upload that institution-specific protocols, and it would say you need to click here. Um, all right, not if it's in, uh, your secured environment. So if it's, so for us, they would have to either get. Chat GPT and use the app or have an equivalent technology in so when you're Microsoft, you're in a secured server so you can put anything in it. We could put chat GPT if as long as it's in that secured server as long as you're using just like you're using Haiku, the Epic app or whatever, as long as you're using the secured app, um, yeah. Um, but that's a good question. Other comments, questions? Any other area where? What are we doing on time? We have, all right, what, any other topics you guys wanna hit on? Yeah, yeah, some of these tools like the prompt engineering or the prompt questions, what is your thought about that because For example, if you go very far one way or very far in a wrong, right or wrong direction based on those initial questions that you're asking your brain. Totally. So, great question. Prompt engineering is the idea of the skill of learning how to prompt. I sent my fellow to prompt engineering class, um, and she learned it. Now, I'm getting to say something like totally cocky here. I don't think you need it. I think, or maybe you need it, but my philosophy on how I prompt engineer. Is I ask it in a way that people think, like they don't think to ask it as this clearly, like, I talked to it like it's a human next to me, and the more. Explicit you are and not just think it's gonna figure out what you're saying, the better your responses and, and it, um, I guess that's what prompt engineering class would do. I also follow people. On TikTok or Instagram that are prompt engineering experts, and each day they're like, all right, here, and you could save those prompts. So, you have like a library of prompts based on, realistically, I don't go to those, I just learned the method of Forcing it out of its rut and saying, hold on. Stop. Um, I don't think you're understanding, like back up or whatever. I talked to it. Um, I don't know. I'm not explaining this well, but it's what you want you you don't want, OK, great. I love what you just said. So number 11 prompt engineering tip, don't give it 5 things at once. Give it one at a time. So if you're gonna say like, I wanna wanna first do this, then I'm gonna want this and this, don't do that. Go, here is what I'm gonna wanna do. Tell it, be very open and transparent. The more you tell it, the better. The more you ask its advice on how to get the result you want. Like, here's the result I want. Help me through this. First, I wanna do this. Once we get that answer, then I, like, that's how explicit I'm, then we're gonna go to stage two. Do you have prompt engineering tips? Not, I mean, besides more like an overall structure of what I'm trying to get, but I told you. The more information you can give it in the structure that you want, the outcome. Yeah, I think if I were to look at people that have trouble and I look at what they wrote, they were, they said way too little. Like, they just assumed it would know and the more exhaustive, I, I'll tell you how I do it. I don't type I use the microphone and just talk. And I'm like, all right, so here's the story. I have this, and I, you know, I just talk and it's like minutes long. That's how I feel like the more I give it, the better chance it's gonna give me the result I want. Um, Yeah, yeah, yeah. Hi Debbie Downing nationwide. So could you touch a little bit upon using AI for after you write your own abstracts and your papers for polishing and how it's OK. I think I put something in. I'm like, wow, this looks beautiful, but just how what. Such a good question. So let me tell you that I am not on the same page with Elsevier, OK. So Witt and I are debating this, and he agrees with what he agrees, and they put out a policy. You cannot use AI and I said, that's crazy. Let me tell you why I think that's insanity. You're trying to get their knowledge explained in a way that makes the most sense to people, that's written well. Why do you care how their syntax is? You want the message to be told internationally. If that's the case, allow them to use word crafting skills, cause it makes it more easy to understand. I'm, I'm an outlier there. I am bullish on that. OK. I submitted. Uh, I wanted to put my dad up for a Wikipedia page. We sent in an AI generated summary of his accomplishments. They immediately wrote back and said, you're rejected because you used AI. OK, so. You're asking the wrong guy, cause I always use it, OK, cause it will always write better than me, but here's what I do, that's a little tricky. I have it write me a draft. And here are the tricks. Number 1, make a draft and edit it. 2. Tell it, do not make this at all sound like AI. Put incorrect syn incorrect grammar. Put some, like, spell things wrong, whatever it is you need to do so that no one can tell. Then I take it and I put it in cloud. And I say, was this written by AI or a human? And then I'll go back and forth until both say this is definitely written by a human. OK, that is not something I should be saying in public, but that is how I do it. Yeah, I cheat 100% because I think that's stupid. We're getting efficient. What they want from me is like, I did, we did a study about these kids that this happened. We found this result. Why do they care how well I put the sentences together, you know. What's that? If you are buying papers, they say that. I know, but I don't know, probably half of you in this room were on this chat email thread with Yama, and we were debating about they do appendectomies always as an interval. They never operate for acute appendicitis. And we were all debating, it's like 20 of us, and every response Yama gave was perfectly written. And he said, I'm using AI to do this. And it made the conversation so much better. And he could take that and have it translate to him. So, I, Sorry, I'm on a, I'm, I'm, I'm my, uh, what's it called? But I, I, I'm a, I, I get, what? Yeah, but I get it. I it's plagiarized. They consider that a form of of. In a fraud. In authentic work But Who did, like, someone must disagree in this room, so we need a contrarian opinion. Nobody or no. Well, Pat, what's the equivalent of saying you're gonna have to do all the math in your data volume. Where do you draw the line? Is that what you're saying? Or, oh yeah, I, exactly. So where do they draw the line? Why? What about spell check? They're OK with us using spell check. I don't know. I mean, obviously, they're afraid that people are literally doing no work. They upload data and they say, write me the paper. We don't know, it's so new, we're all trying to figure it out, but that's a great. Point, yeah. Same. Yeah. You know, I used to always similar things The thing that I was, I guess, heard back was, you know, when you're doing that, it's always like one answer, whereas when it comes to English or when it comes to writing, um, there's a lot of different ways of getting your thoughts and I guess your points across. So I guess maybe that's where that, I guess the line lies where for math there's just one accurate answer, just one answer that you can get, whereas with English there's a lot of different ways to, I guess, present. But you're welcoming wrong answers, which seems crazy to me. I know. Well, it's, so, let me ask you a question. You just made me think of something. Maybe this is the answer. So you're saying, if you give 100% letting the, the, the, besides the data, letting a machine write the story. The story has evolved. It's like a different person's story now, and, and what if? Elsevier had an AI platform that they let us use. Meaning, it, it had restrictions, like you have to put in something to show that it is your thought, but it just finesses it to be, maybe, maybe we could propose that to them. But I hear your point and I definitely always take extreme opinions. So, I like that you challenge that back and I think we need to figure this out. But every everything has to and I mean, if you, who has kids in school? I mean, this is a big thing right now with the kids in school, like, There are actually tricks that the teachers can do, like, they can tell, um. Yeah, different, uh, what is your opinion, uh, on using these tools at the future international face to face Congress? Yes, yeah, because, for example, in Latin 50% of the people don't speak English in Asia is the same from it and, uh, in my mind imagine that my phone, my headphone, and my glasses. My glasses translated. I don't know. Yes, OK, first, I would say that you're not allowed to because Carolina came to the United States to learn English to stay with us, and after two weeks left to go live in a Spanish speaking house because she didn't want to learn the English, um. Um, so, yes, so, so, we were in Japan last week, and the AirPods, you can put them in your ears now, and it translates live as they're talking, it speaks in that language. We should, Carolina, here's what I would propose. OK, for iPEG, let's do it. So it can't just be, it's gotta be so easy that it's really that easy. So, um, it's basically the same as what we do now where you put on the headphones and they have a live translator, but it's just digital translation. So you put on the headphones, you pick your language. In YouTube, I know it's so easy, but I, we have to, we're always the last to do everything. Yeah, we could come up with a way using free software where you just choose which channel, but then when you talk, you should be able to freely speak in your language. Not in having to get them into English, which unfairly has made a lot of presentations look bad because they were trying their best to translate into English. Now, some, so controversial, but I think that would be a big breakthrough at IPEG. That I would put to you that if you're a more eloquent speaker than me, you have the same idea, you just express the same thing. She's not good at the language, so she's not gonna be as eloquent as you. You and I might both speak English. You're more eloquent in getting the idea across. What's wrong with that? Something Ross, you punished if you a better speaker. That's true. I But this is what's so cool is that this is creating controversies we've never had to deal with before. And I I think we're finding our place right now, and these are the questions that we're going to have to. Because you're right, I mean, um, All right. Any topic? OK. Media. Oh, why did that happen? Did I do it wrong? Hold on. I know, but I may have uh. I may have uh done it wrong. Hold on. To help me? Hold on. Let's do this. Um, Let's see. What's that? How do I get my cursor here? I knew this would happen here. Um, I go to media. Why does this not work? If you're Yeah, OK, sorry. So, um, all right. Who's using Notebook LM? OK. Remember I told you, there's the must-haves, must-have. What did I say just as we were walking down here? I said, this is the greatest app ever made. Notebook LM is free. It is a Google product, and it started off like, oh, it's a cute little Google product. They keep beating everybody because they're putting their best tech into it. Notebook LM, I don't know how they're doing this for free, because it uses a ton of working power, but every week it's getting better. So Notebook LM is the idea that it's a workbook. You can upload anything you want, and it can do so much with the stuff, OK? So you have all these workbooks that you have, and they. You can act upon it now. So, to give you an example, um. I, yo, what it is an app, it's a website, but I, there's also an app. Now the problem is the app is not as good as the website, and my kids use it, and it's very limited in the app of what you can do compared to the website. I'm hoping that will change. They change every week, but it is a free app that every one of us should be using. Um, so the idea is you upload content into it, into a, into a workbook. And then you can let it search to add content to it if you want. You can upload anything you want, a video, a PowerPoint, text, whatever. And um then wait till you see the stuff you could do with the stuff you upload. So, I took an example here is I took. I made an executive summary of uh my company and I said, and I did that in Chachi PT. OK. Then here's what so I went to Chachi PT and I said, based on everything you understand about my company, Globalcast MD and its platform, please give me an executive summary. So it went to the internet. It also went through all my history. But what's interesting is you see it went to all these other sources, Google Drive, documents, etc. Because I gave it access to that. You can give chat GPT access to all copy and then I went over to Notebook LM and hit create. Now when you go there to Notebook LM, you can upload any documents you want, give it links, or what I chose to do here is paste the text. All right. So once you paste the text and add all of your other documents, it thinks, and then you can chat with all the documents you uploaded and ask questions about what you uploaded. Or go to studio, you can make a podcast, a video, a mind map, reports, flashcards, and quizzes. This is incredible for studying, for asking questions about a specific list of documents that you gave instead of the internet. So, that video was made, I don't know, a couple of months ago. It has way more things. I just, I'll show you, um, medicine, you need the right knowledge right now, but, uh, traditionally, maybe the best protocol is buried in some huge PDF or like a super long lecture series, not practical, impractical, yeah. Globalcasts promise then is to standardize that gold standard information and make it searchable. Instantly. That's the idea. And OK, to do that, they built the company on, well, two core engines, yeah, complimentary ones. Yeah, so in 2 minutes, you could put anything you want. Yup, you can upload your research paper and it will go, all right, we're diving in today on this incredibly cool study that was done out of the Cleveland Clinic, and it looks at blah, blah, blah. And you could give it some instructions about what to have it focused on, da, da da. Like, we were trying to do it on one of our technologies that's a lighting company, and it didn't focus on the effect on the eyes. So I said, go back and do this again, but focus on the effect on Um, on near-sightedness, and it did it. And the only limitation of, like, I know, Mark Levitt was like, incredibly excited. He's been doing all these things on this, and the only problem I have with it is it's the same two voices. So, everyone knows when you release it, this is a Notebook LM podcast. 11 Labs is another company that you can use, that uses different voices. Uh, there's ways to adjust it, but it's still an incredible way to explain something. So my daughter, Josie, when she was studying her history stuff, she uploaded her notes, and it told it to her in a story, instead of having her read the textbook, which is a better way for her to learn. You can also have him make a movie for you. So when I'm gonna, I give a presentation, it's, it's so much, it's getting so much better now, like, we'll look back and see how basic this is when you see what it's gonna look like in a few months cause I've seen previews. But this is the video it made for me also. So I clicked video, audio, I clicked them all. What if the next big medical breakthrough? Through isn't a new drug, but a better business model. Sounds kind of crazy, right? Well, today we're digging into a company that's trying to do exactly that. Their mission to unlock vital medical knowledge that's been trapped behind closed doors for far, far too long. Seriously, just think about that for a second. It's a pretty startling idea. What if the most brilliant medical innovations are happening in one hospital, but the rest of the world never gets to learn from them? What if our most vital knowledge is just stuck? And this isn't some hypo. It just makes these slides, but there's the slides are getting so good now, and you can adjust it, so it's like a really good movie now. And you could upload. So again, my kids use it to study, so they upload their notes and it tells them the story of the revolution. It tells them the story of whatever. So the teachers get mad, but I'm saying, why? Like, they're learning so much better. I learned from the History Channel because they made it engaging. So why are we opposed to engaging? So, all right, so you want examples of how you can use AI right now in your life. I just used AI image and video generation to help create a PowerPoint slide. Check this out. So in this slide, I want to explain that academic medical centers are now moving to media for education, but to tell the story better, I wanted images. Usually I would just go search on Google and never find the exact image. This time I said, To Nano Banana and Gemini, please create me an image of a boring medical lecture. And then I uploaded that to VO3 and said, make this into a video of this guy giving a boring medical lecture. As you can see from the data, the incidence of postoperative infection remains a significant concern. It's pretty funny. You can see the people even getting up and leaving. Then I said, now create an image of how summary. Videos are much better and more enjoyable. It made a woman watching summary video of a lecture that's more engaging. I uploaded that image to VO3. We can see these cells migrating along these pathways, making their way to the target organ. This is like the new model of how we're going to be telling our story and making PowerPoints and presentations that are much more engaging. So, if you happen to be someone that um I image storytell. I don't usually like words on my slides, I put an image that tells the story. This is how you do it. It just, you describe it, it makes the image, you can then say, turn this now into a video and it makes it into a video. Um, so, creating PowerPoints. Um, I'm gonna skip this one. This one, I made a while ago, it's kind of already outdated. If you go into chat GPT, one of the things, you hit the plus button, one of the choices it has is something called agent mode. We talked about it earlier. That's when you could say, get me a reservation. Remember, agent mode does things for you. So you could say, make me a PowerPoint, and it literally goes and it codes and makes you a really good PowerPoint. You could do it without agent mode. And say make me a PowerPoint just using regular chat GPT, but it's gonna be just like black and white words, whereas if you give it the agent, it'll make you an engaging with images and stuff like that. So agent mode is getting better, um, so chat GPT can make you PowerPoints. By the way, I made this video on an airplane can make you you couldn't hear like using AI software. I removed the background noise if it could create me a professional engaging PowerPoint for my workshop. That has placeholders for videos or images to demonstrate. I'll just skip it because it's, you'll see better and beautiful AI. Has anyone used this? Same thing. You upload your data, it makes you slides. It's very good. Probably the best PowerPoint generator right now, it's called Gamma. So, get gamma, who's used gamma? Yeah, it's real, you like it? Awesome. Yeah, it's really good and it's, it's editable like find me in 6 months. I'll have 6 more programs for you, but right now Gamma is really good. It's gonna get better, um. So, I just made this video, but now. All right, so AI is getting really good at helping to create PowerPoint. There's a bunch of apps that do it. So let's compare a few. So my colleague M took one article, this one disseminating innovations in healthcare, and we had several different technologies try to create a PowerPoint from it. Let's see how they did. So first she went to chat GPT and said, create a deck from this PDF. This is not agent mode. And then it went ahead and did it, and it even asked, what do you want to do next? What type of presentation? And this is what it looked like. Basically, a pretty basic presentation just with words. Um, There's better ways to do this on chat GPT, like if you use the um agentic mode, but this is not bad. Next, you tried gamma, which is really meant for this. You just basically upload information, you tell it what you want out of it. And then it asks you some more detailed questions. And uh you just You just generate it. And you can see this is significantly better than chat GPT. It's got images, it puts timelines, it really does a better job. And gamma is pretty good. You can edit the PowerPoint. It's really good. And then back to Old Faithful, Notebook LM keeps getting better. You just uploaded everything to Notebook LM and tell it to make a presentation. And this is what it looks like. So, it's really incredible. Uh, basically got images, tables, um, it's really graphs. It's phenomenal. Uh, the only problem here is that it only puts it out as a PDF, which is not true, because you, it puts it out as a PDF and then it gives you the choice to, um, To convert it to a PowerPoint. But it's really good as a starting point. You're not gonna do your whole talk this way, but it's good to make some slides. Um. Those slides I showed this week were made that way. Um, infographics, you can all make infographics now. Gemini and Notebook OM and actually, Uh, I'm trying to think for time here. I'll show. Here's another AI tool that is so easy that even a non surgeon can do it. You can make an infographic about anything in seconds. Watch this. Take any document. So here's an executive summary that I had Chat GPT make of my company, and I go to Google Gemini. And then down here, I click on Canvas. Then I paste my executive summary, then I tell it to summarize this in the form of an infographic. And then after maybe about 30 seconds to 1 minute, And there you go. Pretty. So, oh, did I not, um, oh man, I, I made a video for a non-techie person that compared three different infographic makers, and I had them fighting with each other. The, the one that, um, Canva, have you heard of the app called Canva using the AI function. It gives you four choices. So it doesn't just give you an infograph. So, here's four, which one do you like? And then you can literally go in and edit. So, Canva won the battle, as it stands now in January 2026. Um, I will tell you, all of you need to go back. Your organizations need to have someone. Updating you every week. There needs to be an update because, and the organization needs to have people that can go and say, can you show me how to make a PowerPoint? We need to have, that's a new role that I think will be, we're worried for my kids, like, OK, what's the future-proof job for my kids? Like, what should they go into? Maybe this is one of those, like, be someone that can go facilitate organizations and just helping them through day to day. Um, this one is awesome. For those who edits videos here, does anyone edit videos? This has been my favorite tool. I used it already 3 times today. Instead of editing the video, you edit the words. Like, like, like a Microsoft Word and it changes the video. So, it's so easy to use, you could put, and it's got AI in it, so you can upload all the videos and say, make me a video that's such and such length. You can. Watch from a non-techie person, here's a tech you should probably know about. If you edit videos with narration, you should be using DScript. Upload a video with narration, and it shows you the transcript on the side, and you can edit the transcript like a document, and it changes the video. So if you cut words out, it cuts it out of the video. You can even add words and using AI, it adds the words with your voice and video. Here is the text, and here's a clip. And this is the one I'm really excited about because With all of the AI automation tools, with vibe coding, with all the things we have really underpotentialized in the organization, I'm gonna cut it out at vibe coding here, and then I'm gonna go right to this, and this is what it sounds like now. And this is the one I'm really excited about because With all of the AI automation tools, with vibe coding, we need to elevate the workforce to be applying AI automation. So there you go. So it is awesome, and it gets way better than this. I mean, this is the most basic, but it can add titles for you. You just talk, it's called Underlord. You talk to its little AI underlord and say, make this, do this. I would get the script. Um, oh, here's my fight. Oh no, the oxygen, the greatest image generator. I, um. Let's see. Pressure is dropping fast. Increase the oxygen. The greatest image generation software has just been released called Nano Banana, and everyone should have this on their computer. Here are inside Nano Banana, and it's actually called Gemini 2.5 Flash. So when you search for this on the internet, make sure you don't go to one of the fake nano bananas down to the sliders at the bottom, and you'll click on nano banana. That's how you create images. There's also other options like creating videos, etc. but we'll come back to that later. Then you just describe exactly what you want the image to be. And here I'm typing that I want an image of surgeons worried in an operating room. And after seriously just seconds, it produces the image, and you can edit it and keep modifying it. Then you go to the bottom sliders and select video and go ahead and upload that image you just created, and then I give it instructions to make these surgeons worried in the operating room that something is wrong, and then in seconds it makes your video with sound. Losing him. Pressure is dropping fast. Increase the oxygen. And you can upload any of your pictures. So here's a picture of my daughter shadowing my wife in the operating room. So I said, have these two surgeons say, OK, let's operate on this patient and walk over and start operating together. OK, let's go operate on this patient. So it's like an actual gotta give it a try. I think it's time to um. This is just, so Sora, has anyone heard of Sora? Sora is Chachi PT's version of this. It's fun. I don't think it's as helpful, but it's, I think this may be the best video I've seen from Sora too. This is what happens when you take a hilarious doctor. And mix it with Sora too. This is Brad Soboleski, who's a colleague of mine at Cincinnati Children's. Check this video out. What is this? It's a time machine, and it looks like it still works. I can set the date. I could bring help to people long before modern medicine. I've got antibiotics, bandages, everything. Let's do some good. All right, who here has syphilis? Hank, would you care for some penicillin? It's the first line treatment. Hey, Flo, maybe these will help. Thank you. FDR meet Jonas Salk. I think you guys have a lot to talk about. What? Here, try these mood stabilizers and maybe some art therapy and see where that gets you. How many of ye ladies has scurvy? What? Oh, I must be in the future. I better get back to my own time. Looks like I've got some work to do there. Um, you know, this is the best I could do and. You're dancing with my dog. Different, uh, level of, uh, quality. Um, this is what I was just saying like if you're a company and you have stock images you can turn them into videos, but if you don't have a lot of stock footage for does anyone here make but you have stock images you can convert those to videos now but using I had an image and then we, so I took one of the stock photos we work from a company hospitals, they have their stock images, but they don't have videos, so we make videos and I didn't give its more than that and here it did it. So it was a still photo and I just made it into a video. Um, 11 labs can make voice, so if you ever wanna add voice to something, it could be your voice, um. Um, you have expertise the world should know about procedures, research, innovations, but it's often invisible. A few. Hi, welcome to Todd's AI workshop. I stopped using this because it's um It seems like fraudulent. I, I don't know. Like, if I put a video of me talking, it, it seems like I'm pulling one over on people. So if I do it, I would write like AI created, um, voice I do use. I don't know. I feel like that's something you have expertise the world should know about you can procedures, research, innovation, but it's often invisible. A few people might hear it in a lecture or read it on your website. But the rest of the world institutions, hospitals, academics, but you could choose who you want to talk. All right. Uh, what do you guys wanna do? Study, um, automation, medical. Automation, OK. There we go. All right. The idea of automation is gonna get much better, but it's very hard for us in hospitals because you have to use the secured system of your institution. We have a group that we meet every week and we call it agent nerds where we just ask each other, how did you, so, um, the, my project manager is trying to automate. For, so, um, I'm trying to figure out how to do better as a manager, cause I'm not good at being a manager. When I have the different teams, how do I know every week what they're doing without bothering them to send me updates every week. So what I said was, I want an automated system where they are keeping track of their own stuff, but if anything changes in one, in this particular column, which is the update column. It sends me a notification when there's been a change. That way, they don't have to do anything, but I get only what I need to see instead of. A huge Excel sheet. So, this is an example of an automation I made for work, um. So, these are automation tools. Basically, it's getting really good. Zapier or Make are probably. The, the best ones, but most of us who are in a Microsoft institution have to use Power Automate. That's what you're allowed to use. If you go to Power Automate of your hospital, if you go to the Power Automate website, there's hundreds of thousands of pre-made automations that you can use. Like, go check my email, if it's this, then send me this, or you, you'll see there's like thousands of automations. Um, and Basically, it literally just ties apps together. So, if you take something from here and put it to here, it automatically does it. Um, and this is the one I was talking about that um was, we made for my wife, that the patients come in and um it would then make an email, it would, um, it would do all the different things that, I don't know what it did, died, but It does all these different stepwise things that she needs in her office. These are examples of ones you can find online. There are thousands that are there. Um, And I will tell you the automation tool that I use like crazy is Apple Shortcuts. We all have them, we don't use them. So, um, I, you can see how many I have. I, Here's another example of a specific Apple shortcut I use that you may find useful. So my daughter Josie, she's my middle daughter, she oversleeps almost every day, and her alarm clock goes off and she still sleeps. So I need to figure out a way that I can know for sure that she's awake in the morning. So I have the shortcut. Send me an alert when her bathroom light doesn't turn on by a certain time. Check on Josie. I know that she's not awake, and then it sends, and then it sends her an alert, an alarm, and tells me that she's still sleeping. Here's another example, OK, um. I made this one. I've been trying to figure out a way to access voice mode of chat GPT hands-free when I'm driving. This is how you do it. So you create a shortcut and all you do is load voice mode of chat GPT and title it whatever you want, and then you just use the voice. Hey Siri, call chat GPT. Hey, how's it going? Hey there, it's going pretty well. Um, and then this is just about the reminders app I use as a project management tool, so we'll skip this. Um, this one I just posted and people, um, liked it, so I put it in here, but This is how I think every day I have all these things in my head and I always forget them. So, so my wife says that the reason that ADHD has helped me in my career instead of hurt me is because I'm always looking for hacks to manage it. And here's one of those hacks. The problem is I have meetings all day long and tons of ideas all day long. If I just wrote them all down, they'd be a flurry of ideas and it would be Hard to go through. I want them to be organized when I write them down. So I created a shortcut hack that I pushed the button and it says, What's your idea? What's the category, and what's the priority? Then it sorts it into my reminders, gives it a tag, and I divide those into MIT and LIT, most important tasks or less important tasks. It takes me 3 seconds and it's all categorized. Follow my subs. All right. Remember, this is another must-have. Granola AI is a note-taking app. But you can talk with it during the meeting when you're not paying attention like me, so. It's meant, that's not how it's meant to be used, that's how I use it. But really, you type as it's going, and those are the things that it knows to focus on, but it gives you the summary, but it considers what you wrote, but it has these things called recipes that during the meeting, you can say like, I wasn't paying attention, what did they just say? Make me sound smart. What should I be asking right now? What it, give me the summary. Like I use this like if you're All over the place. This is good. This is the best note-taking app if you have ADHD or if you're always thinking of a lot of ideas or get distracted easily. It's called Granola AI. First of all, it links to your calendar. With every meeting, when you go to join meeting, you can automatically just open the note and it ends at the end of every meeting. It'll auto title the meeting. It'll say who was there. As you're recording the meeting, and if there's something important that happens. You just start writing your own notes. It knows that those were the important things for you that you wanted more details on. Here's the best part. So it's called recipes. These are quick questions to ask during the meeting that people would love to be able to do with a live transcription. What did I miss? I use this all the time. So basically, I'm in a meeting and I zoned out briefly. What if, I should, but that you got the gist. So granola AI, um. Let's do. With the notes. Oh, that's a good question. I don't reference them. I don't. I use them when I need to because they have the title of the meeting. I save them, but here's, I'll give you an example. So probably sharing way too much, but I was in a meeting, um, where it was with our CFO and my VP, and they're both way more knowledgeable about the financial details of things, and we're talking about the structure of our new venture fund. And it got into such nitty-gritty, that was way above my head. And I was like, trying to follow, it's all about taxes and, and all the different nuances of where to put the fund and the structure of the fund, and so, I can go back now with that note and chat with it forever. I can say, all right, but what do they mean by this? Explain this structure to me. What do I need to consider? So, I use it to go back after the meeting to talk about it. Um, Does that make sense? But I don't, 99% of them, I don't go back and look at, but I have them forever if I ever want. So it's more for me to just go back. Um, Oh, I see what I did wrong last time. All right, um. I think this is the last thing too, so, um. Study mode, this is more for your kids. Um, here's a tech that probably everyone should know about the study mode in chat GPT. Rather than giving you an answer, it can teach you a topic and it was built by educators. Go into chat GPT and when you click the plus sign, go into study mode, and then pick a topic. And then it will start by giving you the basics. It will give you questions. This is great for your kids as they're studying for classes. It's something we should be. Wait, oh, sorry. Sorry. So this is another simple thing that everyone can do and people are probably already doing is I created a custom GPT for my kids for tutoring. Here's what it looks like. So, I just went into chat GPT and created a custom GPT and gave it instructions. I wanted a patient, encouraging AI tutor that helps students think through schoolwork without ever giving them answers. It guides reasoning, explains concepts, and promotes curiosity and confidence. I had a lot of very specific rules that I gave it to help it guide my kids knowing how they learn. And it doesn't give them any answers. It's just. All right, so this is notebook. So I took pictures of all my daughter's handwritten notes and uploaded the chat GPT to get them transcribed. And I took that transcription and pasted it into Notebook LM as a source. And here Here's what you see. Then it summarized it, and then, I went over to studio, and I made all sorts of study materials for her. So the first thing I did is I went over to flashcards, and it asks you a question and it gives you an answer. And if you want more explanation, you just click explain, and you can chat about it with the guide. Then I created quizzes, and it asks you a question, you get, you can pick an answer. And if you, you can even ask for a hint. And if you get it right, it tells you a good job, but watch, if you get it wrong, then it tells you you got it wrong, but it also tells you the right answer and, and you can even learn more about it. You can even make videos and here I'm playing the video at double speed. All right, let's pull back the curtain on the most complex and honestly the most amazing system in your entire body. So what is this thing exactly? The easiest way to think about it is like your body's own biological high-speed internet. Think of it like a company's command structure. First, you've got the central nervous system or CNS. This is headquarters, the head office. Then you can have it even make a podcast, right, from your DNA's instructions through how your nerves fire. OK, let's jump straight into maybe the biggest debate of them all, nature versus nurture. The classic. On one side, nature, that's your heredity, the genetic hand you were dealt, and on the other, nurture. Now watch this. This is incredible. You can actually interrupt the podcast and ask the podcaster's question. Basically everything else, right? Environmental. Oh hey there, what's up? So what exactly is hereditary? What does that mean? That is a fantastic question, a great clarification to ask right now. OK, let's unhap this definition first. When we talk about heredity or nature, we mean the genetic or predisposed characteristics, essentially the biological units of instruction you receive. And then finally, you can make mind maps that sort of organize the thoughts about what this is all about. So, uh, the kids love it, and they've, I've shown them all these tools. This is the one they gravitate gravitate to. I think that's it, I can just show a minute on agents, but I think we hit, oh yeah, we did medical, right? No, we'll do, we, we can do, uh, let's do medical, but I think we did everything cause I showed you. Um, everything you need. Open evidence. He said you're all using it 100% on the USMLE. What is this and why is everybody using it? Open evidence was released in 2021. Since then, it has become the fastest growing adopted technology by clinicians in history. We've been all looking for non-hallucination, accurate answers to clinical questions. Finally, we have it. It finds their answers through a rag model which is not just searching the web, it's finding verified, updated journal to date that have been published. Everything is cited at the end so you can make sure there's no hallucinations. 40% of clinicians worldwide use open evidence and now. Individual clinicians and hospitals, this is agreements with open evidence to be HIPAA compliant and integrate with your medical records. It now offers CME credit. We're really excited about this and are starting to use it more and more, and we will be doing a workshop diving deep. So here's the thing, it's only US right now, I think. So you have to have an NPI number, but we heard from. The nurses in Akron, they, they said you can send in your nursing certificate, and they will give them the ability to use it, but right now it's only US for the, no, anyone can use it basic, but not, yeah. You are, and you put in, what, so, when Carlos Colonga presented this, he said that no one could use it except US. So maybe either changed or did you, did they ask for your NPI number? So we have to find out, maybe they opened it for everyone now, which would be incredible. That was the biggest complaint from everybody. Um. But if you incorporate it into your hospital, if you have an agreement with them, it is amazing cause it can considers all of your patients. You upload your patient and it says what gives you homegoing instructions, it gives you all these things. Um. You can ask all patient handouts, risk scores, um, different languages, ask for evidence. It's everything if you have that agreement. That's how they make money. Um, this was health GPT we just talked about. Chat GPT Health is coming out soon, and as a surgeon, I can tell you, I'm really excited about this. So much so that I actually tried to make this on my own for my mom about 6 months ago. This is what I tried to build and this is what it's gonna do. So I made this thing called Jackie Advisor, where my mom could upload all her medical records and chat with it and have it remind her about what's going on, what medications is she on, what are her next appointments, because sometimes she'll forget some of these things. And this is what Chat PT Health is gonna do. It's gonna allow you to upload medical records, all of your Apple Health or my. What I would say is go on, look it up, and just add to the waitlist so you get on the waitlist, because right now you have to get on the waitlist. I wanna show you my, I already showed you guys this, um. I think is that everything we didn't do did we do? Yeah, we did. Did we do agents? OK, yeah, agents. Well, actually we didn't, but it's not worth it. That's the thing where I showed you where you can make reservations and stuff like that. So, um, what things did we not talk about that we would wanna hit on? Or did everything get covered? Or is there anyone using something I didn't mention planning to to be an advisor, personal advisor. Yeah. Yeah, that's what, that's what, uh, we're so, actually, to be totally honest. It's, we need that, right? So, I have called like 5 or 6 people. I was like, like, we need to be able to provide like, uh, uh, uh people that can provide support for individuals or for um hospitals. I don't know how to do it yet, but Um, Yes, I think we just, otherwise, we're gonna be, I don't know how to keep up to date. I don't know if someone has a better idea on how to do it, but um. Yeah, what's your framework to integrate this is all fantastic and. I think I use like 1/8 of what you mentioned but. Without kind of sitting there and figuring it out, which takes a few days or longer, you have to kind of like stay involved, otherwise those skills that you figured out go away. What's the best way to have these things, yeah. It's this, I think, I mean, my wife left, but I would, she would tell you, it's. Uh, it's taken over my life. Like, I, every free minute, I am trying to learn this stuff, cause I can't do it during the day. I have no time, right? So, it's like, when do I learn this stuff? How do I play? How do I try this stuff? Cause I have to, for me, I have to. I'm like, this is a wave that is coming, there's no choice here. It's like not learning the internet. So, I'm, I have, I know I have to do it. But I told her this morning, what I'm trying to do is build a team where I have someone else that can help me. That would update me every week, that would do this for me. I wanna do enough where I can learn it, but now going forward, I don't have the time. I would love to have that. So, here's what I would say. I, we talked about it at Steve's Hospital and then a few other places. What we could do is like, what if we do. Like, pick a topic, vibe coding, and we do like a class 11 hour a week or something. We all come in and we, I don't know, maybe it's like a one hour a week thing, or, I don't know. But I don't think it's, unless you get, what I suggested to um one of the hospitals was that they Give protected time to some of their clinicians to be the expert on this. It's kind of like they did that with Epic, like the super users. Maybe there's AI super users in every hospital. Um, I don't, I don't know, um, but I think it's gonna be a, um, I don't think we have time to do it ourselves, yeah. Is that what you're asking? It is. So it's almost going the way of like a statistician where you as the researcher have an idea of what you're looking at, and then the statistician is, you got it. I think it's the idea of a statistician, yes. Except the difference is, what I think is gonna happen over time is you'll need other people less, because it's gonna be democratized to the point where you aren't gonna need. That because it's gonna be so easy to do it on your own. It's gonna be no different than spell checking a document, I think. And the other thing you can do with AI that we didn't talk about was Growth of visibility on the Internet. Like, that's So my, so Diana now, if you type in chat GPT or Google or Gemini and ask about who's the best facial plastic surgeon, she'll come up number one. Like that's another way we're using AI is. Is using ways of understanding how they're finding information. Do you know, like Chachi BT, do you know where it gets its information from mostly? Not Google, cause they know that they're not gonna be able to use it. From Reddit and from Wikipedia, those are the two highest sources, and Bing. And so, if you play the game right. It's unfair, but there are ways to play the game to show up. Um, so that's another big use that hospitals are using AI right now. So the, so the problem with Reddit, cause I would like to, cause I know that if we could get on Reddit. Um, we would show up higher for everything. I mean, at Cincinnati Children's, for example, if I could get on Reddit for every specialty, we would show up higher in every single thing. And there are tricks that we've talked about already that we can do right now. Reddit's hard because you have to be on all the time, responding, chatting, commenting. That takes a lot of work and I don't think you can hire someone to do that. The videos and all that to do it for me, I know you could. Um, But, Rebecca, I, it's, that's the question that's in my mind. When you see all these things and I see how um under-potentialized we all are, it's because we don't have time. So what's the solution for lack of time? And I think the answer is like having a statistician, you need your AI team. Anything else? All right. Well, that was fantastic. Thank you so much, yeah, um. Yeah, I Well, so you all, we have about, the meeting's gonna start in about 1 hour and 15 minutes, so you have some time to take a break. Get refreshed, come on back and we'll get the um the app. Will be I'll say this in the meeting, but just so you know, I sort of made an executive decision because. The snow is not great that rather than.
Click "Show Transcript" to view the full transcription (91214 characters)
Comments