Blockchains and AI: The Path to Superintelligence

Hello and welcome. My name is Mikko Alasarala, the founder and CEO of Inbot, which is a blockchain AI company based in Berlin, Germany. I have also been speaking in international conferences about 2 things. Emotionally intelligent AIs, so artificial intelligences that manipulate us, and blockchain AIs. We won with InBot the global blockchain competition in March in Singapore in the category of blockchain AIs, so it gave a little bit of credibility on my story on blockchain AIs. Today I'm going to talk about first how we are heading towards an artificial superintelligence and then how blockchains will help us get there. Superintelligence as a term means an intelligence that is beyond human intelligence. Check this one. And I'll start with one of the world-leading experts on superintelligence, called Nick Bostrom. Nick Bostrom is a professor at the University of Oxford. He's, I think, a Swedish-born researcher who wrote the book, a very popular book about superintelligence, which is, of course, the name of the book as well. And he says this, and this is very important for us to understand, because we humans tend to be proud of ourselves and also think highly about our abilities, The thing that Nick says is that far from being the smartest possible biological species, we just happen to be the first species that was able to create a technological civilization. So we developed ability to communicate with each other via language, and that enabled us to develop skills during our lifetimes, because before that, the only way we evolved was via the survival of the fittest of evolution. So we filled this niche on this planet, and now we are the dominating species, but it doesn't mean that we represent the best possible intelligence out there. Another guy who I also think highly of is a guy named Ray Kurzweil. He invented a lot of things over the years. He also is the inventor of the concept of singularity, which means that by merging with technology, it is possible to extend human life to infinity eventually. I'm not sure if I agree with his notion of the singularity in general, but there's a lot of ideas that he has that a lot of people agree with, including this idea that there's no inherent barriers for us to be able to develop an artificial intelligence that captures all the essential skills that we have, and all the essentials essential ideas of how intelligence works, and how consciousness works, and all the other things that make us an intelligent species in general. And it doesn't have to be the same, it doesn't have to represent the same kind of intelligence that we represent. That's a very important thing to understand. So when you see all this news about AI taking over, let's say, a game of Go, or AI taking over the ability of diagnosing diseases from doctors, or when the AI is better at analyzing loan papers at the banks, all of these things that are happening, it's very easy for us to settle into this idea that, well, but it doesn't have a consciousness, or well, it cannot do this, it cannot do that, and then therefore we are better. And the most common way of belittling AIs for people, especially for those guys who are working on machine learning, is that, hey, I see how my algorithm is working. I just developed this algorithm, I set in the goal, and it achieved the goal, but it only worked in this very narrow realm of the goal that I set that to do. So for example, if the AI has an ability of discovering pictures of cats from the internet, that was just one of the goals. But one of the things that we don't see necessarily in this picture is that every single day there are multiple tasks, multiple things that we used to be the best at on this planet that we lose to artificial intelligence. So today, maybe let's say there's 5 or 6 or 7 or maybe 25 different things that we used to be the best at, and now AI is better than us at. And we keep doing that on a daily basis for years and years and years and years. All of these AIs are narrow AIs, but as a whole, they start having a skill set, like all of these AIs together have a skill set that is way beyond the human ability. I'm doing this slide to basically show how bad we actually are at all of these things that we think we are good at. So if you compare humans with artificial intelligence, starting from the fact that how we ingest information, how we perceive the world, how we interact, and so on, our hearing, vision, and touch, you know, 3 of the main senses, I mean, there's of course more senses than that, but, you know, like those are the ways that we interact with the world, are actually quite limited. You know, we are able to hear only from a short distance. We are able to only see within the visible light spectrum, you know, within a very narrow field of view. We only can touch with our own body, you know, like, so there's all these different things that we only can do in a very close proximity. And we also are— we learn things like now that you're listening to me, you're ingesting information at approximately 125 words per minute rate. And that's the ability at which we can ingest new information. It's also the reason why it takes for a human child almost 20 years to develop a skill, like via all this education, to become a specialist professional at something. So you go to school at 6 years old, and then you come out as a master's degree from the university at maybe 25, maybe 26, depending on how long you spend at the university. It takes a very, very, very long time to educate this person. It's a very expensive process as well. How long does it take for artificial intelligence to develop a similar skill set for a specific profession? Once we have the algorithms to do that profession, it will be seconds. So now you start comparing the rate at which our computers in our heads are able to ingest information and you compare that to the processing power of the computers, you start realizing that it's actually not that hard to replicate human skill set and then improve on it. Also, the fact that we are sporting all of these smartphones all over the world, like I think there's 2.8 billion of them right now on the field, the AIs of the world have access to all of the sensors on them. So, all the smartphones have a camera, which is the same as vision. They have a microphone, which is the same as hearing. And they have a touchscreen, which is basically able to interact with people via the sense of touch. And all of these things, and a lot of them, are actually monitoring us at all times without us knowing. There's plenty of different interested parties in collecting all the information around us. Including the telecom carriers, including the phone manufacturers, the operating system makers, the app makers, and so on. Like, it depends on how many app makers you gave the right to listen to your microphone or use your camera and so on. It's a well-known fact that a lot of these bigger app makers like Facebook have been listening for the microphones when you let them do that. So yeah, there's a lot of information that is being captured on a daily basis on a global scale. Via these sensors. And this sensor network, for an artificial intelligence that runs in the cloud, let's say that the AI servers on Facebook can practically see the whole world and hear the whole world at the same time. So then you compare that to an individual human, you know, we only communicate via these words, you know, and we are only able to communicate relatively slowly. You know, there's no contest. Like, it's a completely different game. that we are talking about when we call these AIs. Oftentimes people then say, well, but AI is just an algorithm, it doesn't have a consciousness, and consciousness is a uniquely human thing. Well, it's not necessarily so. There's a big question whether consciousness at all is even beneficial to us. According to Max Tegmark, who wrote a very, very good book called Life 3.0, He's an AI professor at the University of Massachusetts, like MIT, and in this book he said that the human consciousness evolved because of the evolution that required that to happen, but 95% of the time our consciousness is directing us to the wrong direction. 95% of the time we are wrong about our analysis of the situation. Why is that? It's because we are biased. And those biases that we evolve over the years, like coming from our upbringing, our culture, or anything that we ingest over these years, is going to cloud our judgment. And this bias makes our consciousness not necessarily so useful. I mean, of course we have it, but we have to understand that it also has an effect on us. So now if you want to build an artificial consciousness, you may want to build it a little bit differently. You want to make it into an objective analytics engine for your existing architecture of AIs under it. So if you have machine learning systems for specific narrow tasks, you just want to build a systemic AI on top of it that analyzes its own doing. And then probably analyzes also its own goals. And tries to harmonize that across all these different potentially conflicting goals. We do have the risk, which is very real risk, that the AIs that are starting to reach to this independent status, so you build a complex enough system that it can start making its own decisions, there's a risk that it inherits human traits because it's being built by human programmers. And right now, the fact that vast majority of AI developers are male, and mostly even white male, there's a huge risk that the AIs of the future are biased against women, against other cultures. So we have to be aware of that. It's not necessarily coming from anybody intentionally making a racist AI, but when people are building an AI that looks like themselves, the thinking or the way that it makes decisions will reflect its own creator. In the same way as we want our kids to be like us, it's just like a very common thing for us as parents to think that we want to have kids that behave like us because we kind of think highly about ourselves and we think that we are making the right decisions. So in the same way, the AIs can evolve like that. Of course, it's possible to mitigate that in multiple ways, But it's important to understand here that human consciousness is not necessarily such a unique thing that it cannot be reproduced with artificial intelligence. So let's go to the curve. So superintelligence, which means an intelligence that is beyond human intelligence, is already here in many areas. So artificial intelligence can best us in a lot of different tasks. I mean, it's better than us in, for example, diagnosing diseases. It's better than us at playing a game of chess or Go. It's better than us in a myriad of other things that are currently out there. So the question is that when is it better than us in general, like in everything? And now we come into, of course, to some extent philosophical discussion on when do you think that something is better than us? Because we as humans behave in a way that Hod Lipson, who is an AI professor at the University of Columbia, I just had a discussion with him at the conference in Germany about these narrow AIs coming up. He said that one of the things that we humans have a flaw on is that we think that the superintelligence will emerge as some kind of Frankenstein, like as the one thing that actually is better than us. And he said that's not how it will work. How it will work is that we will have a lot of these narrow AIs, so a lot of these small AIs that solve one problem, rising up every day, and every time there's a little bit bigger narrow AI coming up that actually seemingly threatens us, let's say a self-driving car, Then we say that, yeah, but it's not the same as us people. I mean, it's basically like we keep ignoring it until it's better than us in everything. And it's going to be always the reaction, it's like, yeah, but it's better than— we are better than this, and we are better than that. So this is the very human reaction of the development of technology. We always find a way of excusing ourselves from these different things. So now, if we go in the exponential path of Moore's Law, and we do not have any hiccups on this process, we will see the superintelligence happening in approximately 2060. Some of us may be even alive at that time. So now if you look at it from that angle, that's basically the path that we're on if we do not have any hiccups. What Nick Bostrom says in his book Superintelligence is that we will probably have some hiccups, and we will have some situations where we run into problems that cannot be solved with software. And even if the hardware and the infrastructure is enabling us to build a superintelligence, we are just not able to build efficient enough systems to represent that in the software. And he thinks that the most likely scenario is that it will happen somewhere between 2100 and 2130. I mean, that's still relatively close. I mean, if you look at the whole history of life on Earth and everything else, it's a very, very, very short timeframe. And then sometimes I ask people who are very skeptical about AI and have this opinion that humans are always better than everything and we are representing the highest level of intelligence in the universe and all kinds of these hubris-filled ideas of ourselves. I was like, okay, so do you think that Do you think that the development of technology will stop? And the answer is that no, of course it will continue. I said, okay, let's agree that it's continuing at the same rate as it is today. So it slows down from this exponential curve into the same rate as it's doing today. And they said, okay, that's fair. And I said, okay, we will still see superintelligence in 2200 or 2300. So we are not talking about very, very long timeframes when the superintelligence comes, no matter how we look at it. So yeah, this is the starting point. So the starting point is that superintelligence is coming. It's coming not as one big thing, but it's coming in the million of small things. Like every day we have some tasks that go away from humans to the AI until there's no tasks left. 50% of human jobs will be lost way, way, way before we hit superintelligence. Because an average human and the most intelligent human, the difference is quite big. So how we look at these tasks and jobs that we have available, a lot of us are doing jobs that are relatively narrow in scope, doing some things that are not needing a very broad set of creative skills, and all of those jobs are under threat. And people may be thinking that the blue-collar jobs of the factory workers and all the other things are the first ones to go, but that's not true. The first ones to go for AI are the ones that do not need any machines. So the AIs that only run on software. So that means the specialist experts working as lawyers and doctors and accountants and so on. So all of these guys are going to be the first guys to be disrupted by AI. I'll give you an example. There's a company in the United States called DoNotPay. This is a very interesting example. It's founded by a 21-year-old guy called Joshua Browder. He's the son of Bill Browder, who is the founder of Hermitage Capital that has had a little bit of fights with Putin, you know, in the past. You may have seen some lawyers dropping from balconies in Moscow and so on. So there has been— Scandals around that. But this Joshua Browder guy, I mean, interesting name for the company, DoNotPay. So what does it mean? Like, he started from parking tickets. He said, like, let's do an AI lawyer. Parking tickets are the kind of legal stuff that no lawyer will touch. I mean, there's no value in it. Like, maybe you have to pay like €60, €80, maybe €100 depending on the city. So he said, like, I have an app, you click a button when you have a parking ticket and you dispute that parking ticket. And that's it. You do nothing, it's completely free, you can file a complaint. And he did that in the US. People started getting millions and millions in parking fines back. He actually started winning more than 50% of all of those cases because of the AI lawyer. So over time, it became insanely efficient in fighting parking tickets. Of course, now you think about it, like for the officials, when you have a very, very credible, good AI-based, like writing on contesting all these parking tickets, like you better just like say, okay, fine, I'm not gonna ask for the money because it's too much fight for me to work on it. So it's basically like people giving up faster because the work is too much compared to the value of the fight. So he won that category. Basically people say, okay, if I get the parking ticket, I'll just go and Press do not pay, and I don't pay it. What happens next? The guy says, OK, let's go to the small claims. Small claims are basically legal disputes that are so small that again, most people don't want to use a lawyer because the value of this fight is basically less than the lawyer fees, and lawyers want to make very, very high fees per hour, as you can understand. So yeah, he goes on that and says, OK, File any kind of small claims court in the United States and we'll fight for you. I mean, let our AI lawyer do that for you. It's still free. It's completely free. Now you go and you want to fight something in the court, you have an app for that. You press a button and the AI lawyer will file a small claims court filing for you. He keeps winning over 50% of the cases again. Okay, so now we have an app that starts winning over half of all the small claims court actions. That's currently an ongoing process. So how do you think it's going to evolve from here? My thinking is, and this is my, of course, my forecast on how that company will evolve, is that the next thing that will happen is that he will go to the middle-class cases. Now we are finally at the turf where the lawyers are being used. And what's going to happen is that he's most likely going to put the system saying, you know, I'm not asking for any retainers. We are not asking for any fees. The only thing we want is 30% of the outcome, for example. I mean, contingency fee. I mean, how many of these middle-class cases will capitulate from the existing human lawyers into AI at this point if the track record of the AI is to win over 50% of the cases? And it already, by the way, has an insanely massive case database from the existing cases that it already won and used for the machine learning. So now you start thinking like, okay, it's most likely what's gonna happen is that all of these middle-class cases will capitulate to AI in no time. And then lawyers will move up on the field into the higher-value cases where they're more complex and more hard to understand. Well, by the way, at the same time the AI is learning about all the possible cases on the middle class, which is a very, very large amount of potential legal cases that it's going through, doesn't take a long time for it to keep going up on the pie. So now finally, eventually, the lawyers are only representing, the human lawyers are representing in the very most complex cases in something and everything else is done by AI. That's how it will happen. This is a progression of AI taking over a whole profession piece by piece, starting from the cases which are the least relevant for humans and then going all the way up to the point that there's no humans needed. And then that's how I see a lot of these different industries will go into capitulate to AI. So in case of doctors, in case of accountants, in case of other things, there are regulatory hurdles for this. So there are some existing regulations that require by law you to use a human to do these things. Of course, that's bad because when you have, like for example in Germany, There's this thing called notaries. You have the notaries mandated by law for every agreement. What the notaries do is that they read you the agreement aloud, and you have to listen to them reading the agreement, and then they charge you a percentage of the whole deal value that you made, whatever the deal value is. You found a company with the game PayHalf, for example, you pay €1,300 for this. It's a legally mandated fee for these guys. You have to sit down for 2 hours, then listening to them speaking to you some boring legal crap, and then you have to walk away, and then you pay them €1,300. It feels incredibly annoying because it's a torture to be there, and if your attention anyhow fades, they are required to get your attention back. So you have to be like watching and listening this thing going on for the whole time until the whole agreement paper is read through. So now, I mean, of course in many countries that's not necessary, but you know, when you create this kind of regulations, then you have to go through them no matter what. So now, of course we have to get rid of those regulations for the AI to take over, but I think AI will take over all of those tasks that we do not regulate away from its reach. And if you look at the data currently available in the world, it's been doubling every 2 years approximately. So since 2017, and in the beginning of 2019, we will have doubled the whole information available in the whole history of the Earth in 2 years. So if you think about this as a progression of speed, by 2021, we have double the information available that we ever had in the history of humanity. And if you look at it from that angle, this amount of data is so staggering that there's no way us humans can make any sense of it or try to do anything about it. So we are using increasingly algorithms to go through this data and make some predictions for us, also make some pattern recognition. also structure and categorize it, make it useful, but it also can make decisions directly from the data as well. So now, having covered the spectrum of the artificial intelligence side of this speech, I'm gonna tell you how blockchains can enable this artificial intelligence to become autonomous. So first of all, in order for artificial intelligence to be able to make its own decisions, it will have to have an ability of spending and earning money. Because spending and earning money in this economic system amounts to being able to make economic decisions by itself. So if you want the AI to be independent, the AI has to be able to control its own wallet. Well, the AI cannot go today to a bank and open a bank account. I mean, it has to be a human who has to do a KYC as a human to this bank account and then control. Of course, it's possible for you to create an AI that then connects over APIs to the bank of a human bank account and then try to transact with that. But it actually has a lot of different types of limitations as well as a lot of human processes that cannot be overcome by algorithms very easily that relates to this. So this is something that can be done with blockchains. So AI can open an account autonomously. It can open as many accounts as it wants in any blockchains out there. It's completely programmatic process. It can also keep its own private keys and everything safe, you know, so it doesn't require any banks, anything. So now you have an ability for an AI to transact financial transactions without any human supervision. This has been actually tried already with this, they call the DAOs. You know, you have algorithmic organizations that are basically incentivizing groups of humans to do things. But this is basically what can happen with the blockchains. So now there's 4 different things that I'm listing here that become possible for artificial intelligence because of blockchains. One is that they are able to make contracts, legal contracts with people. I mean, how are they going to do that? They're going to program those contracts as smart contracts on the blockchain. So now you, smart contract itself is not a legal contract. It's a factory of legal contracts. It's basically a script that creates contracts out of it. So every transaction that you do with a smart contract creates a contract between 2 parties. So I mean, this is actually what I learned in the German law school. Smart contract event organized by local lawyers. I mean, they're defining what the smart contracts are in Germany now because blockchains are commonly used for lots of things. So smart contracts are not contracts in itself, but they are programs that create contracts. So now these programs can create contracts. So every transaction that you do through a smart contract is a contract itself, but they're not the smart contract. itself is not. So anyway, but it enables now that you can, as an AI, you can create different type of contracts with different type of parties. Those parties can be machines, they can be people, it doesn't matter. All of this can be done programmatically without any human supervision. Cryptocurrencies enable now the AIs to move money freely for different purposes. So let's say that, that, and it means that the AI can earn money too. So now if you have an artificial intelligence that is very, very good at spotting anomalies between different cryptocurrency exchanges on, let's say, Bitcoin exchange rates on different exchanges, and doing buy and sell bids using these arbitrage differences on these different things, it can actually start making itself more money with no human supervision. And if it's very, very efficient at it, it could actually create a huge cash reserve, for spending on other things that are important for the AI, including data. So now this is another thing that will come is the tokenized APIs. Tokenized APIs are basically programmatic interfaces to different data sources and algorithms that the artificial intelligence can use, where they can spend tokens automatically from itself to earn new skills. So now if let's say that the AI doesn't know what's happening in the streets of London and it wants to now gain access to the CCTV cameras in London to know the foot traffic, to for example make some financial investments in that area. It could tap into an API that offers that for money and spend its own money via its own wallets for this purpose, gain access to this data, make a pattern recognition analysis and so on, and then apply this information to for commercial decisions. Also, blockchain enables AIs to manipulate people on a large scale. So now if you look at the current scale of the social games as well as all the social media out there, it's all basically algorithmic addictions. So all of these companies are really, really incentivized Because they make money from addiction. They make money from the fact that you spend time on watching YouTube videos or playing games or whatever it is that you're doing. This time is monetized with advertisements, with virtual goods and so on. And now, these virtual rewards that you get from participating in these things, like in these games, like game currencies or likes on the social networks and other things, they are transforming into actual more tangible coins on the cryptocurrency networks. So now if you think about it, now you can use these coins that you earn from different sources, or this artificial intelligence earns, to incentivize people for any kind of goals that it wants. So now it creates reward, for example, on a, let's say a Bitcoin reward for doing some thing A or thing B. Let's say that it wants a human to do a task for it that it doesn't, cannot do itself, maybe because it needs hands, or maybe because it needs some— human conversation. You just create an incentive like Mechanical Turk in Amazon. You just create an incentive on the network, and then people will take that incentive and do these things. So now you have an ability for artificial intelligence to mobilize large groups of people. It sounds scary, but it's already happening. I mean, and I'm saying because I'm working on a company, InBot, which creates virtual cryptocurrency-based incentives and rewards for people. that are directed by AI. So I'm actually in that space myself. And it's very, very scary to see how easy it is to basically trigger different type of human behavior with virtual incentives like this. In order for AI to be able to tap into these vast resources of human beings on the planet, they use something called the Human API. I mean, the Human API in this case is like a programmable interface to people. And what is that? The Human API is this: it's a smartphone which can trigger your attention at any point of time by vibrating in your pocket when the message comes in, by sending a beep sound when the message comes in, by showing a bubble notification on these apps and showing this notification on the screens to get your attention. I mean, once you get the attention, then you can direct this human behavior into specific goals by using these virtual currency rewards. The only thing you need to do is to get the attention first, drive them into an application on the phone where they see that, hey, you can earn this much money by doing this, for example. And now you have an API to mobilize a very, very large group of people for any goals that you have in your mind. I mean, sometimes these rewards do not have to be even monetary rewards. They can be Rewards of sense of belonging, social exchange, and so on. Like, there's lots of different rewards that drive our human behavior. But all of these are already known by all of these companies out there, like the Googles and Facebooks of the world. They already understand how this works, and they are applying it incredibly efficiently in getting our attention and getting our time as an investment. So now you have an artificial intelligence which has its own wallet in this vast world of blockchains, and it has the ability of driving human behavior. It has the ability of making its own decisions, and it has the ability of earning money and spending money. What does that make them? It makes them autonomous economic agents. So now, once you start having things like this, and this is what I made as a prediction in 2017. I made a prediction that by the end of 2020, we will see the first company that goes from zero to billion in one week. And if you think about this as a speed at which you can build value, it's completely beyond any kind of human comprehension. We cannot understand something like this. And it's so crazy that it's probably going to be decided as a criminal activity or whatever it is, but we cannot, because we can't understand it, we will think it's threatening. And if you look at what happened, like what has happened so far, 2 of the fastest paths from zero to billion has been done so far in the history of humanity in cryptocurrencies. The first one was Binance, you know, the cryptocurrency exchange which is currently the largest in the world. They went from zero to billion dollars in 6 months. And that happened in the course of the last 6 months of 2017. Wow. There was another one, which was now called a criminal activity, a Chinese exchange called Fcoin. They invented a new incentive structure called transaction mining. And this transaction mining pulled in people so fast that they went from $0 to $1 billion in 2 weeks. And they went from $0 to $17 billion in exchange volume in a matter of less than a month. After that, the Chinese government attacked them, and I think some people went to jail or whatever happened. But the thing is that now the benchmark for the zero to billion is 2 weeks. So I think by the end of 2020, the 1-week goal is still possible. And I think it looks more likely to happen than not happen. And I'm quite confident that this will happen in the space of cryptocurrencies because of the ability of mobilizing large masses in a completely algorithmic way. And most likely the business that does that is going to be algorithmic in nature. So if you put any humans in the middle, any kind of human negotiation in the middle, the timeframes will instantly expand to much, much, much longer. So this is the power of blockchains. And I want to bring into attention at this point, because I've been lobbying from my location in Germany, this cryptocurrency taxation thing in Finland. You know, so Finland, Finnish tax administration decided that they want to punish all of the cryptocurrency transactions in the maximum possible way by saying that you cannot deduce your losses. Every transaction, every time the interpretation of the cryptocurrency is made in the worst possible way. So if you are mining cryptocurrency, the currency is a currency at that moment, but when you're spending the currency, it's not the currency, it's an asset. Because every time they just interpret these cryptocurrencies into something that is, from a taxation perspective, the highest value item for the taxman. So now, having this kind of punitive tax scheme, which I think is illegal, but at least it's running, is making it impossible for Finnish companies to participate in this global public blockchain internet of value. And I do think that it's a necessity for any country to be part of this network. No matter what they think now about the history of Bitcoin and other things, doesn't really matter. What matters is that what blockchains really enable us to do, they enable us to have value exchanged between people all over the world without any middlemen, without any gatekeepers, without anything that comes in the middle, completely programmatically if necessary. So now you can have a self-driving car like they do in Germany, self-driving car, that is tapped into blockchains to be able to transact with different, let's say, pay fees to the paid roads, for example, or be able to pay for charging the car in some charging station in the city and so on without you doing nothing. You just basically sit there and all of these transactions are completely automatic and autonomous. So now this network is so important in my opinion that There needs to be public awareness of this topic that Finland cannot screw this up so badly that they decide to opt out. You cannot be opting out from the internet. That's the kind of thing that is happening, in my opinion, right now. It's saying, let's build our own little pond of internet that doesn't have any cryptocurrencies in it, that just has a blockchain, and the government is going to pay for it, and it's gonna be like this small pond for Finnish people that can stay away from the rest of the world. But I think it's a bad decision. So we need to keep talking about it until some of the politicians will understand that this needs to change and it needs to change fast. Because even if Finland wants to be in the forefront of artificial intelligence in the world, they also have to be aware that these AIs that are going to be on the leading edge are going to leverage blockchains for a lot of things because it gives those AIs a lot of autonomy. Coming back to the smart contracts, basically the smart contracts today are made possible by blockchains. So in order for you to create contracts that are stored in a way that it is not controlled by any centralized party, where like I can contract with anyone I want, I need to store this contract in the blockchain. I want to store it as a transaction And it's a hopefully immutable transaction that cannot be mingled with later on. I mean, you can have contracts that are reversible, but it has to be reversible in a way that there's another transaction that reverses the contract's intent on the second transaction. And when you have it structured like this, you have an ability of scaling any business using completely automatic agreement creation by software. Between different parties. And this kind of scaling is unprecedented scaling compared to the current world where you still have human salespeople like talking to other humans and then you have a negotiation happening by words and then you have some kind of agreement drafted by lawyers and then you have something to do. It takes a long time to do this. It takes a lot of money to do this. And it all increases inefficiency in the systems. So it's much, much better that we move into this programmable realm Where all of this can be done by basically creating some kind of preset rules saying that these contracts can happen within this realm. And then you just let them out and let the algorithms do the thing. And coming back still more and more to these tokenized APIs. So the tokenized APIs are also something that can be only done on blockchains. You can of course have paid APIs on different services out there. For example, you can use Stripe Payments or you can use PayPal or you can use something for paying something on the internet, but that actually currently requires a human account. Again, coming back to this thing that as soon as you require something to be owned by humans and controlled by humans and secured by humans, you lose out the ability of doing completely autonomous growth or autonomous behavior or autonomous economic agents that are beyond this human behavior. And I do think that in order for these APIs to work really fast, you need to remove the human negotiations from the angle. So right now, if I want to tap into, let's say, dataset from LinkedIn, I need to go and meet those people and I need to negotiate with them and then they need to agree for their commercial interest to work with me. And then they need to basically draft the contract, and then you have some kind of access to some of the data in some places that they want. And that's a very, very inefficient way of building things, you know, if you want to have fast data growth. So that's why you need these autonomous agents being able to directly contract via tokenized APIs, via smart contracts that are already having preset conditions for your access to this data. so that it can capture and grow faster. And as a final thing, is this running out of time now? Yeah, we are kind of running into the launch button. Yeah, so as a final thing, I've created this draft of how you can create free universal healthcare using AIs and blockchains. So this concept basically has an idea that you can basically take in money using the human greed and human like these biases as a mechanism of funding this system. So what you do is that you create a health fund that sells services to the existing insurance companies, be it the governmental insurance companies or like private insurance companies on the health side, at a lower cost than they're currently getting from the existing doctor services. And because they are going to make more profit as an insurance company by buying these lower-cost services, they're going to be happy to buy these AI-powered services from from this entity. But because this entity doesn't pay any dividends to anyone, it doesn't try to make profit, it just tries to aggregate as big a fund as possible. It will basically be an engine of sucking all the money from the existing insurance systems and then tokenizing it on the backend and then using those tokens for paying for health services for those people who contribute data to the actual AI. So now you can have a mechanism where people are providing incentivized with autonomous reward mechanisms to contribute disease data as well as their personal health data into the blockchain-based database that then is researched and understood and pattern recognized and everything is machine learned by AI, which then becomes a better doctor that can then provide better services that can then monetize from the existing insurance systems until the whole existing insurance system is dead. And this is how you can come into a place where everybody can have a free healthcare service eventually funded by the existing system originally. And that's the end of it. Thank you for your attention and time. Thank you.