Well, thank you everyone for being here. I'm very excited to share some of our insights, answer some questions if I can, and just keep it a very collaborative session. So just stop me at any time if something doesn't make sense. If you want to add to the conversation, if you want to ask questions, we'll be very happy to take those. As I'm sure most of you have come across the very rapid pace at which artificial intelligence is taking over the mainstream, and it's taking over the mindshare of people both in the media and in real life. And us as a firm, we specialize in this type of disruptive technology. So we obviously spend a lot of time trying to make sense of it. And this technology, just like any other before, moves at a very rapid pace. So trying to define what we understand as intelligent technologies as a category, we try to put those into three main categories. The first one is artificial intelligence, but it expands to things like robotics and neural networks. Don't worry about the complexity of maybe some of the words here. They're not that difficult to understand if you draw some analogies. So artificial intelligence is just trying to make systems actually intelligent and smart. Most of the times, 10, 20 years ago, when a system felt smart as you interacted with it, it wasn't actually smart in the background. It just had a lot of if-then statements in the background. So humans had sat down and said, here's a decision tree. If somebody says this, do this. If somebody says that, do that. And as I'm sure you all can understand, that wasn't a very smart system in the background, but it felt relatively smart when you interacted with it. What's different now is by using models like neural networks that have been around forever. Neural network technology has been run for about 50 years. But given the data and the compute capacity that we have today, neural networks can actually be smart in the background as well. So instead of giving it if-then statements, the way that neural networks are trained is you give them a lot of examples. So you say, this is a question, and this is the right answer. And if you give them a million of those questions and answers, the neural network itself starts to pick up what the pattern and the logic and the intelligence is. So the network, that's why it's called neural networks. It's because it learns information the way the neurons learn information in our brains. But it's not very complicated. It's like a child. If you give it enough good examples, the child picks up what you're trying to teach it. Robotics is obvious, but it is the physical manifestation of these intelligent systems that can take that logic in the background and actually physically do something with it. And so if you break down any intelligent system, it's going to have three major components. It's going to have context, which is its memory, what it knows. It's going to have reasoning, which is the logic and the reasoning that it takes from that memory. And it's going to have actions, what it can actually do. So can it send an email? Can it push a car, instruct a car to go forward? But all systems are as simple as that. It's the context, the reasoning, and the actions. All intelligent systems are built like that. Now, you must be wondering why there's so much more discussion about this technology now than there used to be before. Because technology has been around forever. And that's because of the slide on your left. It's because in 1956, even though a lot of these concepts existed, it was cost prohibitive to use them in any intelligent systems. Because one terabyte of memory cost $100 trillion to have. Right? And today, one terabyte of memory costs $100, right? So it wasn't that humans have not been trying to solve this problem for a long time. It was that the infrastructure, the physical constraints didn't allow them to solve this problem before. And these numbers are getting even more, moving down even more aggressively now that there's so much capital, both financial and intellectual capital, going into trying to solve the problem of memory, of processing, and things like that. But AI is not just possible today, AI is necessary. And it's because of the chart on the right. So if you look at the bottom, that's the price of things over time from 1997 to 2023. The only things that have gone down in price are technology. Because technology is inherently deflationary. It becomes cheaper over time because of the chart on the left. But all the other things that don't have today a lot of technology are becoming prohibitively expensive. Like college, which I'm sure anyone with kids knows. And medical care, household energy, and stuff like that. So it's important to look at these charts because you can apply technology to many of those use cases and actually have a real shot at bringing the cost of those down. Because in the trajectory that they're going, college, healthcare, there's not really another reliable, logical way to bring the cost of those down other than applying a lot of technology to it. So what if you could teach 1,000 students, not with 100 teachers, but you could teach them with one teacher? You would expand the access to that education to a lot more people, but you would bring the cost down just as much. One of the crazy statistics is that at Stanford, there's just as many teachers and administrative staff as there are students, which is ridiculous, right? You might as well give everybody one butler, right, and a professor to walk around with them. But that's why the cost of education is so high. And it's not getting lower. There's no real credible path of it getting lower. Healthcare, as we all know, is medical expenses are the number one reason that Americans go bankrupt. And that's not getting cheaper either, right? But the only way to make a real dent in many of those things is through the application of technology. And we actually have, as a society, a real chance of making all these things accessible to the vast majority of people and make them much cheaper if we apply technology. The other big shift that's happening is even in the perception of the experts and the scientists that work on this stuff, and it's happening very rapidly. So artificial general intelligence, AGI, is what you think about, what do you see in movies when you see the Terminator or something like that? That's what artificial general intelligence is. It's a fully autonomous, fully intelligent machine that can be as impactful, as powerful as a human. And what the experts thought just in 2021 was that it would take about 40 years for us to get to AGI, a Terminator type of, Skynet type of being. But as we've seen the development of the technology move so rapidly, they now expect it to happen in about 2025, 2026. And that's a really rapid change because before, we could all kind of feel comfortable in our world, knowing that it was probably 20, 30 years away but now, it's probably two or three years away. So we're not talking about this technology taking decades to change the world. People, most of the experts now expect it to happen over the next couple of years. And that's both scary and also should be exciting if you are on the right side and if you're embracing the technology. The other thing that's very obvious to see is even with small improvements in the model, how much of a, how rapid of an improvement you see on their performance. So what you see on the left side are standardized tests like the SAT, the LSAT, the GRE. And the blue bar that you see is a newer model and the gray bar you see is the older model. And you can see that just a model that took maybe, there's a difference of about six to 12 months between these models. You can see how much more effective the blue model is compared to the gray model. And effectiveness to the point where on an LSAT, this model can score in 80%, in the 80th percentile. On the GRE, it can score to the 99th percentile. So very quickly, the things that are very difficult for humans to do, computers are able to do them quite well and in quite sophisticated fashion. There are also some very complicated problems like the AB problem that people thought would not, computers are not smart enough to handle those, but they've already started handling those. And one thing I will tell you because I spend a considerable amount of my time on this stuff is, whatever you think the capabilities of the models are or whatever models you see out in the open, there are models that are at least a 10x improvement of those already being developed and ready to be pushed out. And that's a constant thing. Every time we see a new model, within about six months, we see a 10x improvement. So we're not talking about a small five to 10% improvement over time. We're talking about 10x improvement, step function improvement every single time. Any questions, comments to add here before we keep kind of moving forward? Okay. Yeah. Yeah, I'm sorry about the formatting at the top here. We don't use PowerPoint, so when we export things, sometimes it gets messed up. We use an AI to build our presentations, actually. And I'm not joking, I'm like serious about that. So it helps us, like this presentation, how many slides is it, 18 slides? I think it took the team about 20 minutes to put it together because of the tools that we use. But we have a pressing issue, and that issue became very obvious during COVID and after COVID. We have a labor shortage, and that labor shortage is only going to get worse as people get more education, people get higher expectations of their lifestyles, at least on the physical labor side, we're seeing a very significant shortage. And there's not really any other way to fill that shortage because we're talking about millions of people. There's only really a few ways that you can solve that problem. And we think that humanoid robots are a very credible path to solving that. And we've seen examples of that already. Before this whole change and how powerful these AIs were, you've seen that the time to ship in an Amazon warehouse went down from about 60 to 65 minutes, down to about 15 minutes because of the Kiva robots that they use. And that was technology from about 10 years ago. So this technology is getting rapidly better. But if you want to solve real life problems that the US is facing and countries all around the world are facing, the only real credible path is to use some version of artificial intelligence. This is the expectation of what humanoid robots are going to look like over the next 10 years or so. You're seeing revenues in the billions, you're seeing deliveries in the millions of robots. And we'll show you some examples of that here as well. I promise I won't keep talking, I'll show you some cool stuff. And the reason it's happening is because the cost of building one of these robots is coming down significantly. So a high spec robot in about 2022 would cost about $250,000 to build. That cost is already down to about 150. And the low spec robots are down to about $30,000. These robots don't take time off, they don't complain. They don't go to sleep, they just keep working. So a 30, even $150,000 robot that's super specialized is far more cost effective and efficient over time. And as I'm sure many of you have already experienced it, ChatGPT was the fastest growing consumer product ever. It got to 100 million users in two months. And the closest thing we'd seen to that was Windows 10, WeChat, TikTok, which were step functions behind it. And this was the first version of ChatGPT. Like I was saying, every single time you see a model that you're using, you should think to yourself that it's going to be the worst model that you'll ever use, because that's just the truth. These things are getting very good very quickly. So I wanted to show you some real life examples of how the technology's being used to solve real problems. These are not experiments, these are out there in real life. The videos weren't working in the presentation, so I'm just going to jump onto YouTube. This will be more fun than me talking. Oh, I'm sorry. So you cannot see that. There we go. So the first one is a company called Axon. They're the ones who created the Taser. They're really focused on trying to make law enforcement less lethal. And this is another one of the products that they recently launched. 13th, 18th Avenue South coming separate at the neighbors at 290 to 11. I'm just pointing over. You're going a little faster. The school's not okay. It's 25 miles an hour. You're going 35. I'm just pointing over. You're going 35. So, that's the example I'm going to show you. So, that's example number one. Another example, I don't know if you've come across a company called Klarna. It's a buy now, pay later company, mostly in Europe. I don't really like the product itself or the business, but what they've done for the customer service using AI has been pretty incredible. So, they introduced, about three months ago, an AI chatbot. And I know that we've come across AI chatbots at like Amazon customer service and Walmart customer service, and they kind of suck. And so, you kind of always are like, get me to the representative, get me to the representative, right? I'm not talking about chatbots like that, right? Remember what I was saying in the background of some of the intelligent systems before, was just human code. It just said, if this happens, do this. That's not what we're talking about here. What they've done is they've taken, Klarna has taken, because also when you jump on one of these calls, they say, you know, we're recording this for customer service purposes or whatever. They actually took those recordings and said, which ones of these got the highest satisfaction scores, trained the model on that customer service representative, because that's the one that our customers are the happiest with. And so, they took their absolute best customer service reps and used their examples to teach the computer what a good customer service rep looks like. They were able to, this new bot is able to handle 2.3 million conversations in the first two months, which is two-thirds of all of their customer service requests. It's getting the same work done as 700 human agents. It's getting the same satisfaction score as human agents. It is also reducing the time, the wait time, sorry, the resolution time from 11 minutes to two minutes, because you don't have to wait for the bot. It has perfect recall, perfect memory. It doesn't have to go check your account. It already knows what your account is. And so, everything becomes way faster, way more efficient. But the really interesting thing about it is that it can run 24-7, and it can run in 30-plus languages. And I used it myself, because I do speak multiple languages, and I started the conversation in English, switched over to another language, switched over to another language, and it kept doing the conversation with no problems at all. And it responded to me in my other languages, and I could respond back in English. You can have these customer service agents speak hundreds of languages all around the world, even understand specifics about the culture of each place that people are coming from, and it really is an incredible experience. But overall, it's not just about the experience. It's about how much this hits their bottom line. And Klarna expects that they'll save about $40 million this year by implementing this bot. So, really incredible, tangible results. Something that I don't know if most people realize how powerful these customer service applications have already become. Let's look at some Tesla robots. It's impressive to me. Oh, sorry. Yeah. And in typical Tesla fashion, they also made them dance. I forgot about that part. Yeah. But they've come a long way. The robots that you remember, we're not talking about those anymore. This is another interesting one. The people that are building the robots are about to be disrupted. So Devin, some people have come across Devin, I'm sure, is the first AI engineer. Because engineers always felt like they were the last ones to get disrupted and go, because they were the ones disrupting everybody, but that's not the case anymore. Hey, I'm Scott from Cognition AI. And today I'm really excited to introduce you to Devin, the first AI software engineer. Let me show you an example of Devin in action. I'm going to ask Devin to benchmark the performance of LLAMA on a couple of different API providers. From now on, Devin is in the driver's seat. First Devin makes a step-by-step plan of how to tackle the problem. After that, it builds a whole project using all the same tools that a human software engineer would use. Devin has its own command line, its own code editor, and even its own browser. In this case, Devin decides to use the browser to pull up API documentation so that it can read up and learn how to plug into each of these APIs. Here Devin runs into an unexpected error. Devin actually decides to add a debugging print statement, reruns the code with the debugging print statement, and then uses the error in the logs to figure out how to fix the bug. Finally, Devin decides to build and deploy a website with full styling as the visualization. You can see the website here. All of this is possible today because of the advances that we've made in both reasoning and long-term planning. It's a really hard problem, and we've only just started, but we're super excited about the progress that we've made so far. In the meantime, if you'd like to try out Devin on your own real-world tasks, send us a request below and we'd be happy to forward it to Devin. Yeah, and we've been using not Devin, we've been building a Devin for ourselves, and we've taken a sprint. So a sprint is how software teams break up their planning. Our sprints used to be two weeks long, so we would usually handle two or three features in a sprint on applications that we would build or earn the investors in. Our sprints have now become one-day sprints because we're able to start the day and using these tools by the end of the day deliver two or three features. And that's with us messing around with the tech for about six months. It will only get better and better. We expect sprints to become hour-long, where at the beginning of the hour you dream up a feature, and by the end of the hour you have the feature. So software development, any sort of technology development, is going to go through quite a bit of change. Let's switch back to our presentation. And as expected, the market interest is at an all-time high. We are participants in the same market. We invest in AI companies, both public and private, so we see this on a daily basis. But private investing in AI has gone through the roof. The mentions of AI in earnings calls for Russell 3000 companies has gone through the roof, as you would expect. But there's a number of risks, we feel, when it comes to investing in AI or building in AI, and they largely come from the types of companies that will win over time. And we feel that there's only two things that will matter over time, which will be how much proprietary data you have, and how wide your distribution network is. And we can go into a lot of details on that, not super relevant here, but where you're seeing most of the capital flow is either the capital is flowing into venture capital, so very early-stage AI companies, or the capital is flowing into really late-stage big tech companies, so Microsoft, Nvidia, Google, where we see kind of a greenfield opportunity is in the middle, where there's companies that are not considered AI companies, but if you apply AI to their businesses, it will have a transformational impact on the overall company. This is a funny video. At least I think it's funny. So this is the founder of OpenAI, the creators of ChadGBT. Let's see what he has to say. A lot of people think that this is what AI investing needs to look like. It needs to be investing in companies like OpenAI. Whether we burn $500 million a year, or $5 billion, or $50 billion a year, I don't care. I genuinely don't, as long as we can, I think, stay on the trajectory where eventually we create way more value for society than that, and as long as we can figure out a way to pay the bills. Like, we're making AGI. It's going to be expensive. It's totally worth it. Yeah, so he was asked, you know, how much is too much? Like, how much are you willing to spend to build AGI? And basically, his number is unlimited. We don't believe this is the right way to approach this. We believe for most private companies, we are absolutely in a bubble, because the company I just showed you, the software developer, Cognition AI, they just raised almost $200 million at a $3 billion valuation, and they're a six-month-old company. So we absolutely believe that the private side of the market is in a bubble, and I think investors would be right to be very careful there. We also believe that late-stage companies, the Microsofts of the world, AI is going to have an impact on their business, but it's not going to move the needle for them. They're already ginormous businesses. So we also think the focus there is wrong, because at the end of the day, just like every other technology, fundamentals matter. You cannot pay an exuberant price for growth. This is a chart from 1945 to 2020, and the number one thing, as you would expect, that drives stock price over time, or the valuation of companies, is earnings. Everything else does not matter. Growth in earnings matters, but growth for the sake of growth does not matter. And so we feel that there's a lot of capital being thrown around at companies that don't have any clear path of getting to earnings. And we're not doing this for charity, we're doing this to make earnings. But that is it. This is us, if you'd like to get in touch with us. And happy to answer any questions. Yes? Would you expect that there'll be a mix of use of sort of closed systems, where the information that the AI can access is limited by, say, a company or an industry, versus broader open systems? So who are the companies that will help end users and companies develop that? Yeah, so let me repeat the question for the video. It's a great question. It's whether we expect that most of the development here is going to happen in closed systems with access to proprietary data, or it's going to happen in broad systems with access to global data. What we feel is for the vast majority of use cases, especially the use cases that you can charge a lot of money for, there is no point in training those models on Reddit and Twitter data. There's no point in training it on just open source data. Because for the vast majority of use cases, you don't need a large language model. You actually need a small language model that's very specialized to your use case. And that's something that I don't think people appreciate enough, and that's why they keep investing in companies developing large language models, like open AI and things like that. For let's say a law firm, whether their model has an understanding of the discussions on Twitter or not is completely irrelevant. But whether it has a very specialized understanding of the Texas business code is super important. Whether it has an understanding of how they have fought and won cases in the past, or how they have fought and lost cases in the past, is far more important than an understanding what happens on Reddit. And so we feel that proprietary data is one of the two advantages that are necessary to build an AI system that is valuable and that you can charge for. And most companies in the world actually do not have a proprietary data advantage. Law firms have never collected their data in a way that it can be used today. We financial planners have not collected their data in a way that it can be useful today. And so even though they have the ability and the chance to have a data advantage, they have not seen data as a real asset. They've seen data as an expense that they have to deal with over time. And so as a thought experiment, if I were to hire, if I had enough money to hire every single engineer at Tesla, would I be able to recreate full self-driving? And some people would expect that, yes, if you have the same intellectual capital, then you would be able to recreate full self-driving. The truth is that I couldn't even get close. Because the model itself, as I was saying, is about 50 years old. It's free technology. It's available to everybody. But the millions of hours of video data that Tesla has, no other car company in the world has that. And they're still not collecting that data, right? Even though it's so obvious how valuable it is. Because the model, like I said, is completely stupid until you give it enough examples. And Tesla has millions and millions of hours of examples. And they can train that model to be better and better over time, and everybody else is going to either license the data from them or buy the data, or otherwise they have no chance of building a full self-driving solution. But most of the capital and focus is going to companies that are three months old. So they don't have a data advantage, and they don't have a distribution advantage. So we feel that there are going to be some successes there for sure, but that's going to be very low probability of success. So one of the really weird things that will happen, even though we're going into this very high tech world, is that the old school consulting firms will become actually very profitable for a short period of time. And then once they've built all the AI systems, they'll become less profitable. But you actually need these consultants and technologists to come in and build the technology that will then replace them. That's probably where the first place that at least enterprises are going. They don't want large language models. They want small language models. They want companies that understand how to build those, train those, and deploy them. And once those are done, then they've taken their data as an expense item of what they pay to keep the data, and now they've made it intellectual property as an asset on their balance sheet. So there's a number of firms here in town, all over the world, that are specialized in taking your data and making it intellectual property for the first time. So I feel that the use cases where you can make a dramatic impact on the business, where you can charge a lot for that particular application, they cannot possibly come from open source data, because you have no moat when there's an open source data set. But if you have a closed source data set, which will become the most valuable asset on your balance sheet, then you have a real advantage. Yeah. Yes? In the objective, it talks about Web 3.0. I'm not admist it. Can you explain what that is? And before you close, can you tell us what year we die to our general intelligence robots? Yeah. So initially, I had thought, we invest in any frontier markets that are complex to understand. That's what we've specialized in. And when there's a really complex market, we get very excited, because we feel we can make logical, sensible portfolios around it. And so one of the markets that we invest in is what some people call Web 3. So essentially, the way that they think about the internet is Web 1.0 was when the internet was websites that were static. So there would be a newspaper that you could go read. There would be a yellow book, so you could go search different things. And then Web 2.0 was when users started creating the content that went on websites, like Facebook, TikTok, YouTube. These are not static websites. This is where our content is what makes the website. And so that's Web 2.0, where ultimately, there's a few big players that kind of control the whole industry. Web 3.0, according to people, comes from the advent of blockchain technology, where we go back to having these websites controlled by individuals instead of companies, like Facebook and Google. And so that's the overarching theme, is the underlying technology is blockchain to democratize the internet again, because it went from being totally democratic to being super monopolized, and now they want to make it democratic again in Web 3.0. And so we invest across this Web 3.0 space, as well as AI. So that was one of the original topics, but I felt that this was a more relevant topic to what people are experiencing today. And Web 3.0 has actually become quite more mature than it was about 10 years ago when we started working through it. And so we expect in about 10 years, this will be kind of the boring topic as well. And so we'll keep moving forward. I do feel that we have real risks from AGI. Those are totally real risks. If you look at the smartest people in the world, when it comes to AI, they're extremely concerned about AGI. And that's why you continuously see people like Elon Musk signing petitions and doing things about slowing the advent of the technology down. Because I think a lot of people feel that there's a software or hardware limitation to what the tech can do. And especially when there's a hardware limitation, it's just a hard limit. It's like a physical limit. And so you can't blow through it. It takes time to figure physical limitations out because you can't change the laws of physics. Software problems are even a little easier to solve, but still difficult to solve. I think what people don't realize is that it is neither a hardware problem nor a software problem. We have all the hardware we could possibly need to build an AGI. We also have all the software we could possibly need to build an AGI. It is simply a matter of how much data we can feed it. Because when a model is only 80% accurate, it's not because the software isn't good. It's because you haven't given it enough data. If you give it enough data, it will become 100% accurate. And there is no stopping that from happening. The only limitation is how much clean, digestible data we can give it. Because it's not dependent on human logic at all. It's just saying, show me enough examples and I will learn. And if you show it a trillion examples, it will learn every edge case that you can imagine. And so it will get to 100% efficacy as a model. So that's why I think people are so scared. The people who really do this every day. Because it would make me feel a lot better too if they said, oh, there's just not enough processing power in the world. We need to wait 10 years before we get there. I would also be able to go to sleep real comfortable, right? But I know they have enough processing power. They just lack clean data. And they're getting there. And so you feed it enough data and it will become completely intelligent. That's why the experts, remember the chart I was showing? The experts think it's in the next five years. And I wouldn't disagree with them. Yeah. Yeah, absolutely. It's happening. It's happening every day. It's happening in our own business as well. So in the beginning, we would just invest in technology. But then we started thinking to ourselves, you know, how can we go to work every day, see this incredible technology, and not use it to make our investment process better? Because our own investment process was take a piece of paper, write down the stuff, and make up a decision, right? And so we started implementing the same technology for our investment process. So I'll give you an example of how this has at least changed our investment process. So we used to subscribe to all kinds of data providers, and we would try to sift through the information on earnings calls and try to make sense of where we should invest our capital. And it would take one analyst. At most, one of our team members could cover 50 companies decently, right? Because you can't possibly remember and understand 50 companies and their business models, right? And so when we started working on AI as an investment, we started first by extracting every single SEC filing that has ever been made for any company in the world. And we found every single transcript that's ever been transcribed for every company in the world, every presentation. We found every piece of information that you could find about a public company. And we could go through the entire history of a publicly listed company, 20 years, 30 years, 40 years, of every publicly available information about them in about a minute of processing. And if we did that for 10,000 companies, it would be even less. It would be a day. We could go through every single thing that they had. And what we found was that humans generally have a good instinct on what matters, like this. Everybody kind of knows this, right? But I'll tell you like an insight, for example. When people look at P to E ratios, right, price to earnings ratios, everybody uses the earnings from the last 12 months. That's just how all the platforms and everybody builds it, right? But if you run enough data through the model, the model will tell you that what matters actually is the earnings for the next 12 months and the estimates for the next 12 months. And so that was a key insight. And if you put that key insight in here in any of the best stock pickers in the world, their performance will become dramatically better. So from the perspective of like how much information it can absorb and understand and recall, humans obviously can't keep up, right? Because it just learned 30 years of a company's history in a minute. And it has perfect recall. It has perfect logic. Doesn't forget. Doesn't get confused. And humans just can't keep up with that, right? You just can't. And so the best path forward for anybody in finance is to embrace the technology, use it to take the relationship and the care that you have for your clients is probably the only skill set that will stay relevant, right? Everything else you should be able to supercharge your process using the technology. If you run enough simulations of how the markets have done over the last 100 years, it can build you the exact specific model portfolio for every single client at every single age and every single gender and every single state. Like it can figure that out for you. And you don't have to spend any time on that. You spend all of your time building the relationship, understanding their problems, and then use the technology to come up with a proven solution. Because right now what we come up with is a solution that makes sense to us, was taught to us, but none of us can say that it's a proven solution without looking at the data, right? We all look at data, historical data, to make sure that what we're suggesting to our clients has been proven over time. I'm saying you should have an analyst that has perfect recall about everything that has ever happened help you with that. And so I think what will happen is you'll be able to take care of your clients better. You'll be able to do it faster. You'll be able to take care of more clients than what you can today. And I'm seeing it absolutely across the board. All the large firms, they're all using the technology to just be faster, better, cheaper for their clients. Yeah, for sure. Yes? What's your guess about the impact of this on health care and better clinical outcomes? Oh, yeah, that's one of the most exciting things for me because I'm not originally from the U.S. And so when I moved to the U.S., one of the most shocking things to me was this was, you know, this is the strongest economy in the world and people were struggling with medical bills and medical care, right? And so that was just always shocking to me that with all the entrepreneurship and all the capital in the world, the Americans have not figured that piece out. But I think they actually can now. As an example, which some of you must be familiar with, they have completely replaced radiologists with A.I. Because radiology is like perfect use case for the technology because you have millions and millions and millions of scans and you have millions and millions and millions of diagnosis, right, that humans have done. And you also actually know whether the diagnosis was right or the diagnosis was wrong. And if you think about it from that perspective, that's like you should be like, oh, perfect solution, right? Because you give it all of the examples and you give it all of the right answers and you also give it all the answers where the doctors got it wrong. And it will find patterns that no humans have found. And it will say before any other human can understand what it's even looking for, yep, this person has a risk for cancer. And by the time the radiologists pick it up, it's already too late. And so they've completely replaced radiologists and they've taken the time of reading a radiology report and typing it up to like a few seconds. And obviously the cost comes down with that. Everything else comes down with that. This is just one example of one of the things that it can do. And so I think we should all be extremely excited that this technology is coming around because it will help save a lot of lives. And even if it doesn't take your lifespan from 60 years to 80 years, it will take the cost of health care down so significantly where people are not waiting weeks for appointments and spending thousands of dollars that they don't have on these solutions. And another interesting use case was that you do have a lot of patents that make the cost of pharmaceutical drugs very expensive. But that was because the computers weren't powerful enough to come up with thousands of versions of the same drug with enough of a difference that they don't infringe the patent but can be distributed basically for free. And so there's some companies that are working on that. And they're saying, hey, we know what's in this. Let's try to get the same efficacy out of something else with enough of a difference that we don't infringe the patent at all. And so why not have an AI build a pharmaceutical company that gives out drugs for cost without infringing any patents? That would be great for competition, first of all, because I am a capitalist. But it would be great for the cost of health care. Any other questions? Okay. Thank you very much.

artificial intelligence

neural networks

robotics

proprietary data

law enforcement

customer service

healthcare revolution

radiology

investment landscape