Trusted Digital Assistant - "Human OeS"

--- tags: Papers --- # Trusted Digital Assistant - "Human OeS" ## Rise of digital twin for human being ### Digital Twin A digital twin is a virtual representation that serves as the real-time digital counterpart of physical objects or processes. A digital twin is created by collecting enough data (shape, physical properties, material, behavior) to be able to model an object in the digital space with enough accuracy that it can serve a specific purpose. The simple "digital twin" which for sure everyone had a chance to interact with is any type of digital photo of a product (e.g. shoes) that you purchase online. By creating a "digital twin" of the pair of the shoes you can see how they look, you get information about size, materials, etc. You can "experience" the counterpart of the physical product to some extent at a vast distance, remotely without the need to leave the leaving room. Of course, this is a very simplified example just to show the base idea of the digital twin: Physical object -> Data -> Digital Twin Organizations using more sophisticated digital twins for way more complex use cases. Collecting enough data, they can create a digital twin of complicated physical objects, put them into a simulation and check how they perform. Check physical limitation (crash tests), stress test critical components and verify that they behave as designed in different conditions. As you can imagine, this significantly can reduce the costs of such operations. There is a reason why the digital twin approach disrupts entire product lifecycle management (PLM) in almost all industries. It enabled product owners to gain access to an enormous amount of real-time data which does not need to be collected by monitoring and hooking sensors to physical objects but by running their counterparts in simulation. How many times you can afford to crash a Boeing 737 to test how it can handle harsh weather conditions? In a simulation, you can do it all over again without the need to worry about the costs or consequences. The IoT revolution brought a step further where due to low costs of sensors we actually can hook in real-time to physical objects and monitor them as we go. So we are not limited anymore only to the early-stage life cycle of the product in the lab before releasing it to the market, we gained access to the data of the life product used by our consumers. Based on those data we can create their digital twins (not anymore a product digital twin, but a concrete instance of the product) so something which was formerly impossible practice now reaches anyone. How we can use it? Based on the real-time we can now simulate digital twin in a counterpart environment, predict failure, react faster, and inform customers even before the product will be damaged. With a simple controlling mechanism built into the devices we not only can monitor it and inform, but we could remotely fix the issues, enable self-healing functionality when needed. Could have full control over that device without even bothering customers, who could enjoy using the device without the need to ever worry about anything. Digital twin brings us huge potential spanning across the entire product lifecycle, from R&D, design, manufacturing, service and maintenance, and daily operation. ### Digital twin for a living being As explained above digital twin is possible due to the amount of data collected about a physical object which then can be more or less simulated in the digital form serving specific needs. If we can do it about a physical object can we do it for a physical person or animal? Just looking at the data sets that big tech companies are collecting about us we could easily conclude that each of us already has a digital twin in the wild. Not always controlled by us, sometimes even without our consent but it is there. We are already quite well connected and sometimes we give away definitely too much information so companies can leverage that and "run" our digital twin and do profiling on it to see what we would like to watch, buy or eat. Following the concept of the digital twin from physical objects space they don't have to live in the past and see what you did but they can try to predict what and when would happen in some cases they don't event need to predict anymore and they can tell you what you should do. Even though some tech companies collected an enormous amount of the data about you they still struggle quite a lot. Did you try to ask Siri what is your phone number? Or maybe you bought something recently and now getting the ads for the same thing just cheaper in another store? Why even with such huge data sets and having digital twin companies still get it wrong? Is it so hard to understand that: "I already got bought that device and don't want to buy another one even if it would be half price"? It could be a simple answer for that: **economics of correlator**. ### Economics of correlator The correlator is an entity that tries to gather data about you and correlated it with specific behavior or create a prediction. Many of you experience such behavior through the ads industry where collected data allows companies to propose "best" offers and products to us. As we already mentioned that does not work always as intended. The main reason for that is certain limitations related to simple economics. Due to the vast amount of data and different limitations (technical, or economical) getting real-time data with proper context is still not possible. We are not a simple IoT device that measures just 4 data points. We are complex living beings interacting in an even more complex ecosystem. Monitoring all that at scale is no easy task. Not only that, there are economic constraints where extracting real-time data would cost more than the benefits so would not pay off for those companies. Economics of correlator describes a relationship between correlation costs and value which generates spots in which extractable value can be accessed. As we can see in the below diagram, the **costs of the correlation** need to be lower than the **correlation value** so it would be profitable to extract the value from the data. By manipulating mentioned curves we can control the economics around the data, influencing the behavior of the data market. ![](https://i.imgur.com/clRBybu.png) Fig. 1: Economics of Correlator In many cases, the most valuable information is those which are near "Now". So the services can adapt to your current needs. It does not help if you would get a proposition of the product which you already bought. But as we can see on the diagram the cost of extraction grows significantly as soon as it approaches "Now". The reason for that is that collecting data in real-time is very challenging and in many cases requires full surveillance of a person. If we like it or not information is generated not only by us: social platform activities, buying behavior, but as well our friends and family: sharing photos with you on social platforms, sharing your phone number through phone book with the specific instant messaging app. Based on collected data companies can do powerful predictions. You could, for example, infer a social profile of a not yet born child based on the information which her/his mother is posting on a social platform, the country in which he is living in, areas, parent's background, and more. As we learned about physical objects having enough data we can understand how to react and control the device so it can operate in the desired way. If we would apply the same principle to the digital twin of a living being, we could gain similar control. Most people would say immediately that would never allow for that happening but would like to have such insights into a person in your family who is suffering from dementia, or maybe an underage child who is going missing? As with any tool, a digital twin can be used in different ways for good and bad things. The question is how we can make sure that we limit the bad things to the minimum and leverage it for those good things. I think now we all can agree that we already have a digital twin and it is possible to create it for any living being. We just need to face a few challenges like access to the real-time data and who would have control over all that. How do modify the economics of the correlator in a way to have access to real-time data without losing control or compromising privacy? ## Changing rules to solve the problem A digital twin is a very powerful concept that can bring a lot of benefits for physical objects as well as for living beings. The change is to create Digital Twin in a way that we won't be slaves of the system. Shoshana Zuboffin in her book The Age of Surveillance Capitalism provides evidence that already collected data by organizations helping them to predict our future, but they are using it to model our desire and opinion so they don't need to guess anymore, they can tell us what we need and desire. I think everyone would agree that this is not the ecosystem in which we would like to live and we need to find a way to prevent degradation of the system as well revers what already happened. As we learned about the economics of correlators, we can manipulate the data market influencing the costs of the correlation or the correlation value, so that value extraction can happen in desired conditions. E.g. to make value extraction possible to approach "Now" on the timeline and get the data in real-time we could change assumptions about who in that situation is a correlator and drive the costs of the correlation to the minimum. If the data would be collected by the user himself, the costs dramatically drop since he already has devices able to collect data in real-time about what he does and store that on his side. We not only make it affordable because of the scale (just one individual instead of millions) but as well we change the control over the data and since it is for my purposes the consent dramatically changes. If any big company would ask you to collect your medical data without any specific reasons, you probably would never agree with that. But if the medical data would be collected for your purpose under your control so nobody else would see them without your consent that changes the game. Now you become your correlator, costs of correlation drop to the minimum, which allows us to increase the surface of value extraction to get the most out of it. To achieve that we need to equip individuals with proper tools and security so they can perform such activities without worry of being a spy on or losing sensitive information. We need Trusted Digital Assistant. ## Trusted Digital Assistant A trusted Digital Assistant (TDA) is a set of components that allows you to operate in a digital space. It helps you to collect, register and monitor your data. It helps you manage your digital identities and enable you access to digital services with full control over the data flow. TDA becomes you in the digital world. It works on your behalf, responds, and keeps you safe. It is your gateway to the digital world. To understand what TDA is we need to start with "Me" in digital space. We as humans (or any other animal) are not equipped with any hardware or software to interact directly in digital space. We need help, at least for time being. To operate in digital space we need to have some hardware that would allow us to connect to the digital world. Hardware in most cases can't work without software. With such a setup we need to make sure that between Us and the hardware there would be security that would allow us to fully control that device and make sure that we would be able to detect cases if someone would manipulate it in a way that it started acting maliciously. Same we need between Software <-> and Hardware to make sure that the software running on that device is doing exactly what we are expecting and would always be under our control. To achieve such binding we can use biometrics to allow a device to be linked to us as an individual human being, through components like TEE/Secure elements we can assure that this link would not be easy to break, and the same we have to do with software so when it runs we can cryptographically verify that the version which we are running is exactly the one which we are expecting to run, there is no any unwanted modification nor malicious software running on the side. Both hardware and software should be transparent from the perspective of what and how it works so we can assure about a proper characteristic of those components. Since who we are is not only a matter of what we have or what credentials we possess, we need to take into consideration as well 4th element which is the environment. Similar to you walking on the sandy beach you leave a trace in your environment same in real and digital space, everything we do, patterns of our behaviors, connections, and activities can be used to identify us as individuals. This is very important since by nature our identity is decentralized, it is located partly in all the places and people with whom we are interacting. You can think of it as if you would lose your memory, how your family and friends can help you restore who you are. The below diagram summarizes those characteristics of Digital Identity Architecture. ![](https://i.imgur.com/EVnkpfl.png) Fig. 2: Digital Identity Architecture With Digital Identity Architecture in mind, we can design a system (set of components) that allows you to operate in digital space. We call it "HumanOeS". It is an operating system necessary to exist in the digital environment securely. Using HumanOeS we can build our Trusted Digital Assistant which we can trust and rely on since it would be us. It would be us in digital space. With help of TDA we can have: - authentic real-time data - full control over generated data - privacy preserve mechanism to protect and serve the physical counterpart - ## Governance of TDA Same as you are not allowed to use any car on public roads we need to be careful about which TDA can be used as your digital twin. To be able to rely on TDA we need to know how it is built we can think of open hardware approach for the hardware part and open source for the software, to be able to investigate the mechanism how that works, what is inside, does it do what it suppose to do and so on. Those values are very important but the average person is not able to verify that on his own. We need to delegate that to entities that can stand for us and protect us against bad TDA providers and certify TDA components for specific use. The intention is to have as many TDA providers as possible, through fair competition people would be able to choose the version which fits them the best. But to protect individuals from malicious providers we need to have an approval process that can verify the correctness and compatibility of such TDAs in a given use case. Having one centralized organization which would decide what is "THE TDA" obviously won't help much, we need to do better to avoid bias and subjective opinions. This is why a fractal approach for decentralized governance would be needed where different rules and conditions could be applied on different layers. For example at the EU level European commission could decide what type of HW and crypto software is allowed to operate TDA. Health organizations can decide what are requirements to deal with medical data (e.g. which TDA providers are allowed in this industry in the EU, excluding companies that operate outside of EU jurisdiction). Banks could define what are the conditions and rules which TDA needs to fulfill to be used in financial services. This way we could end up with different types of TDAs with different levels of security depending on need or use case. To assure interoperability TDA needs to operate on a common infrastructure which then can be shared among all of those use cases. Similar to building roads for any type of the allowed vehicle we need to build base infrastructure for TDA as well. This would assure that in case a user wants to use a more secure TDA for medical data or financial services he can easily move and switch TDA providers depending on the needs. Same if a person would move from an EU country to Brazil, he would be able to upgrade his TDA to the TDA which is accepted in Brazil based on their rules and requirements.