From Digital Twins to Digital Selves and Beyond

Front matter

Franz Barachini Christian Stary

From Digital Twins to Digital Selves and Beyond

Engineering and Social Models for a Trans-humanist World

Prologue

According to evolutionary theory, the first amino acids evolved from a primordial soup based on inorganic material. They formed the basis for organic life. Since then, several transformation processes of biological life were traversed. Most scientists believe that these transformation processes are not continuous but volatile. Nobody knows when exactly self-awareness evolved. Mankind evolved not steadily but gradually with sudden boosts and more steady phases in between.

Approximately 70,000 years ago, the cognitive revolution started with the development of a language. In contrast to animals, the modern homo sapiens was able to use this language as a means for storytelling. By inventing stories, myth could be created. Yuval Noah Harari (2019) coined the term virtual language for that. This virtual language allowed the creation of common imaginations. Common imaginations and myth formed the cornerstone for the management of large groups and even complete nations.

Concerning the younger history, Karl Jaspers (2010) introduced the term axial ages for extraordinary leaps in mental evolution. During the first axial age between 800 and 200 before Christ, the thinking frames of all higher developed societies were created within a few generations independent from each other. The first woodblock printings were also invented during that period. Later, around 1700 anno Domini, the search for a proof of God’s existence yielded the foundation of natural sciences.

Since the invention of the Turing machine in the year 1936, more and more innovations and inventions were produced in shorter time. The usage of mass products follows exponential curves. This is due to the existence of modern computer systems.

In the middle of social, environmental, and technical upheavals, it is legitimate to ask in which directions humankind will evolve. Overpopulation and as a consequence environmental pollution, loss of biodiversity, and, as a consequence, pandemics and climate change threaten our existence. In the long run, global warming could make organic life on earth extinct. In order to extend human existence, mankind has some possibilities to survive for some time. We either can leave the solar system or we can try to transform ourselves back to inorganic material.

ix x Prologue

Leaving the solar system is only possible with power unit technologies which are not invented yet. We think that the transformation from inorganic material to organic life and back again is also promising. The latter step is known as trans-humanism. Ray Kurzweil (2005) introduced the idea of “uploading” brains on powerful computational substrates. This process and the amalgamation of artificial and real brains are known as singularity.

This book is a tiny contribution to the ideas of the Gilgamesh project. However, in contrast to the Gilgamesh project, which is more oriented toward the exploration of eternal biological life, we think that research initiatives redirecting attention toward capabilities of digital artifacts and their impact on artificial life help redefining common grounds and disparities. In this way, we want to lay ground for informed co-existence and co-creation opportunities, either driven by biological or digital capabilities. Consequently, the book is about deepening the understanding of the relation between Cyber-Physical Systems (CPS) as socio-technical systems and advancing their digital representations in terms of intertwining human and machine intelligence. In exploring this avenue, it is crucial to recognize that digital twins as digital selves should be able to develop emotional behavior, enriching the socio-cognitive habitat of Cyber-Physical Systems.

Why This Book?

Picture a traffic ecosystem including fully autonomous cars among other vehicles. It represents a socio-technical system as some of the cars are driven by humans and some of them are not. Fully autonomous cars do not require human drivers. These vehicles are capable of sensing their environment and are preprogrammed to operate without human intervention. As they do not require human interaction, they are technology artifacts of the socio-technical traffic ecosystem.

From a traffic system perspective, both fully autonomous cars and human-driven cars are equal components. Being quite different with respect to perception, decision-making, and operation control processes, each party has to control itself in traffic and to meet regulatory and mobility requirements. Human-driven cars, semi-automated vehicles, and fully autonomous cars can go anywhere traditionally operated cars can move and should be able to perform tasks experienced human drivers can accomplish.

Autonomous cars have some particularities human-driven ones may not have and do not rely on when it comes to socio-emotional behavior. Technologically, fully autonomous cars need to rely on sensors, actuators, complex algorithms, machine learning systems, and powerful processors to monitor behavior, identify events, plan the next intervention, and run software for behavior execution. Many human sensing capabilities are copied by autonomous cars when creating and maintaining a map of their surroundings. Radar sensors monitor the position of nearby vehicles. Video cameras detect traffic lights, read road signs, track other vehicles, and look for pedestrians. Lidar (light detection and ranging) sensors bounce pulses of light off the car’s surroundings to measure distances, detect road edges, and identify lane markings. Ultrasonic sensors in the wheels detect curbs and other vehicles when parking. Overall, many detectors of environment signals collect data to be processed by technologically relevant (re-)actions.

What if digital processes do not only process sensory inputs or plot a path and send instructions to the car’s actuators, which control acceleration, braking, and steering, but operate beyond these capabilities? What if a fully autonomous car does xi xii Why This Book?

not only rely on rules avoiding obstacles, predicting driving behavior, and recognizing various objects, but rather uses algorithms for predictive modeling and situation recognition to take into account social and emotional patterns of behavior? What if cognitive computing is used to enrich cyber-physical ecosystems beyond following traffic rules and obstacle navigation and develop individual emotions?

Once socially enriched fully autonomous cars become part of the traffic system, socio-cognitive intelligence qualifies them as integrative and socially adaptive parties of this ecosystem. Similar to human-driven cars, autonomous vehicles could include behavior comparable to subtle cues and non-verbal communication. For instance, instead of making direct eye contact with other ecosystem, participants or systems expression of affordance could be encoded relating to politeness. Similar to reading facial expressions and body language of other participants, behavior judgment and activity predictions could become part of autonomous vehicle behavior.

As can be shown by developing real-life scenarios, these mechanisms target connecting to people. However, expecting only for upcoming car generations more than 300 million lines of code (cf. http://www.synopsys.com), digital modeling of autonomous systems becomes an indispensable asset in socio-technical development. Hence, we need to think about encoding socio-emotional intelligence in digital twins and establish them as digital selves. Digital selves are expected to contribute to what could termed “social compatibility” of actors in a continuously transforming digitized world. Key are transparent and executable models, structuring cognitive and social processes for humans kept in the development loop.

Overview

Since digital twins increasingly become high-fidelity digital models of ecosystems and their dynamic evolvement (cf. Durão et al., 2018) in the first part of this book, we provide information on their background and their development in the context of Cyber-Physical Systems (CPS) and their industrial application. We discuss the engineering of so-called humanoid socio-technical settings mimicking cognitive and social skills. This tendency lays ground to build up sovereignty on developing some digital identity or self, finally controlling individual development by digital means. Taking this role, digital twins are real-time models capable of controlling their adaptation and, finally, next-generation development, thus representing digital selves. They become the core of socio-trans-human developments.

In the second part of this book, we introduce relevant approaches to awareness and the emergence of consciousness in artificial agents. We start with discussing the awareness of identity in digital twins as prerequisite to social representations. We then discuss whether social behavior of digital selves can result as an emergent phenomenon. Progress in simulating emotions by using game theoretical approaches is presented.

These findings lead finally to considering the coexistence in terms of a symbiosis of digital twins and models of socio-emotional intelligence (cf. Kiister et al., 2020) enabling digital selves on the background of highly connected ecosystems and increasingly digitized service operations. In order to structure the contributions from the various fields, we adopt a tri-partite framework. It has been developed to capture human-related issues, organizational elements, and technologies including their mutual relations. We adopt this well-established conceptualization for complex system analysis and development—cf. Oppl and Stary (2019) for the digitalization of work (processes) and Ackerman et al. (2018) for the socio-technical transformation of society-relevant systems, such as healthcare.

The adoption of the framework extends the human perspective to individuals in the realm of trans-human developments enabled by technology. The technology xiii xiv Overview

Collective Perspective: Organizing socio-technical entities

Digital Twin

_— Technology Perspective: Cyber-Physical Systems

Individual Perspective: Human = Transhuman Sovereignty

Fig. 1 Digital twin development as socio-technical endeavor perspective recognizes the cyber-physical nature of upcoming ecosystems. The collective perspective includes the socio-technical structuring of individual elements toward a systemic whole (Fig. 1).

In the following, we provide a chapter-by-chapter overview for each part of the book. The first chapter describes major milestones in computer science. These milestones form the basis for the implementation of digital twins and digital selves—the third axial age, still being in progress, started around the 1930s.

Although the idea of the digital twin stems from industrial automation, its objectives have systemic relevance. It ranges from effectiveness in digital transformation to participatory design and transparency of change, thus touching principles and structures of human life and socialization. Consequently, we dedicate the first part of the book to a system and systemic understanding of the reality we perceive and design in the course of intensifying digitalization, in particular when utilizing the full potential of upcoming generations of Cyber-Physical Systems (CPS). We start out reflecting on developments of so-called humanoid socio-technical systems that increasingly mimic human behavior including generating novel behavior and structures of digital intelligence.

We proceed with introducing relevant concepts of systems (engineering). They address the dynamic nature of system, once we zoom into socio-technical systems and need to consider interactions as substantial to humanoid socio-technical systems. In that context, we explain the nature of Complex Adaptive Systems (CAS). Since these systems consist of a variety of actors who can be in control of change and development, the understanding of the concept of System-of-Systems (SoS) helps in capturing concurrent processes under the control of multiple actors.

Since many communities and bodies organize themselves through data-driven events rather than interaction, we analyze the shift from focusing on data to behavior

Overview Xv modeling as target of development. It not only enables a unifying perspective on systems but also features stakeholder participation, in particular through no- or low-coding elements. This shift brings us to models as baseline and frame of reference for development. Once models as digital representations, like in the case of digital twins, become the focus of development, design activities will be intertwined with engineering cyber-physical ecosystems. Thereby role and task understanding is crucial for development and operation. The latter features the emergence of digital identities that can be part of CPS, requiring digital identity management techniques, affecting predictive analytics and learning.

In the second part of the book, we discuss possible incarnations of digital selves. We Start to speculate about the emergence of consciousness in artificial agents. We proceed in describing the research area of agent-based social simulation. We argue for game theory as powerful theory in modeling intra-agent behavior of large crowds. Next, we describe problems of social order and we discuss whether these problems will emerge in the course of cooperating digital selves. We show that this will not necessarily be the case, as it mainly depends on the type of implementation and on advances in computer science.

Subsequently, we present modeling of emotions based on deterministic dynamics for populations arranged in a lattice. We use non-cooperative game theory as a vehicle for social simulation. We proceed by presenting the implementation of the business transaction theory as a means for cooperating on a stochastic basis. We compare the results with the deterministic approach presented previously.

The upgrading of digital twins to digital selves is discussed in Part III of the book. Here, we present a symbiosis of Parts I and II. We show under which preconditions digital selves based on digital twins are able to establish and run digital ecosystems. The final chapter in this part is dedicated to opportunities and modes of creating reflective socio-trans-human systems, exploring mutual control and continuous development.

The epilogue is congenitally speculative in its nature. We present our thoughts on future developments of artificial life in computational substrates.

Acknowledgements We would like to thank Springer for supporting us in organizing the content and production of the book, in particular Ralf Gerstner. We also appreciate the effort of Ines Golubovic for editing and polishing the figures.

Supported by Johannes Kepler Open Access Publishing Fund.

Chapter 1 A) Major Historical Landmarks in Computer ge Science

We consider 1930 as the beginning of a new age. This period is characterized by developments which can be expressed by exponential curves.’ By means of digital computer systems organized in networks, information can be stored, retrieved, and distributed simultaneously to the crowd extremely speedy. This is the precondition for the quick exchange of information between different scientific disciplines and paves the way for exponential growth of ideas, inventions, innovations, and products. Furthermore, each real event, process, or thing can be expressed, modeled, or simulated in virtual environments.

In our mind, three people, namely, Kurt Gödel, Alonzo Church, and Alan Turing, can be identified as godfathers for that period. Alonzo Church (1941) invented the lambda calculus, which can be regarded as the basis for many programming languages far before the first computer system had been built. Alan Turing (1937), who was a scholar of Church, verbalized the somehow complicated formalized findings of Kurt Gödel (1931).Turing replaced the formal language from Gödel by a simple mechanism consisting of a tape divided into cells, a head that can read and write symbols on that tape, and a state register which stores the state. This mechanism models a machine on a piece of paper—later called the Turing machine. The Church-Turing thesis hypothesizes that anything that can be “computed” can be computed by some Turing machine.

In the year 1941, 10 years after Gédel’s publications, Konrad Zuse invented the first programmable Turing-complete computer system Z3. He applied the first high-level programming language, which he named Plankalkiil (Giloi, 1997). For the calculations, the Z3 computer used circuits consisting of relays and vacuum tubes.

The Boolean algebra (Boole, 1854, 2016) became the mathematical basis of digital computing. Claude Shannon (Shannon, 1938) and Victor Shestakov both showed that electronic relays and switches, called gates, can realize the expressions

! Exponential curves are not well understood by humans. In primitive times, humankind was only confronted with linear problems.

© The Author(s) 2022 1 F. Barachini, C. Stary, From Digital Twins to Digital Selves and Beyond, https://doi.org/10.1007/978-3-030-96412-2_1 of Boolean algebra. This was the foundation of practical digital circuit design. Finally, in 1945, John von Neumann proposed a computer system architecture consisting of a memory, a control unit, and an arithmetic logic unit equipped with input and output facilities. This architecture formed the basis for the modern digital computer.

Soon afterward, the bipolar transistor based on silicon replaced vacuum tubes. The next generation of digital computers used integrated circuit chips. The invention of the MOSFET transistor enabled the practical use of MOS transistors as memory cell storage elements. From the year 1966 onward, memory capacity increased by a factor of 10 every fourth year. Following this exponential curve, we can conclude that in the year 2028, the complete storage capacity of an average human brain will be stored on a single memory chip. In order to prove this claim, one only needs to solve the following equation:

In parallel, due to rapid MOSFET scaling, MOS IC chips rapidly increased in complexity at a rate predicted by Moore’s law (Moore, 1965). There is a debate among scientists about the life span of Moore’s law. Due to advances in lithographic techniques, new materials, and 3D wafers, we think that Moore’s observations remain valid for at least another decade. However, at the end, we will push physical limits due to tunnel effects of semiconductors. However, new technologies, such as quantum computer systems, boost computing space and power to higher limits.

In parallel to the exponentially growing hardware advances, there was a need to express algorithms in a more elegant and understandable way.’ Therefore, new programming paradigms and programming languages were invented. During the 1950s, FORTRAN, a language for solving scientific problems, dominated. PL1 and COBOL were applied to commercial problems. During the 1970s, PASCAL and its derivatives dominated the programming scene. In addition to the imperative paradigm, functional and declarative languages, such as LISP and PROLOG, were applied. Around 1980, the object-oriented paradigm prevailed. Languages as ADA, C++, and JAVA, its pendant C#, and finally scripting languages completed the 50 years lasting development process. In the field of software engineering, software development environments and the Unified Modeling Language (UML) as a way to visualize the design of a system became common.

Tim Berners-Lee invented the World Wide Web in the year 1989. This pioneering innovation provided the capability to connect all worldwide operating computer

?The YEAR is the unknown. The left side of the equation represents the number of bits we would need to store the brain capacity. An average brain consists roughly of 10'* neurons. Each neuron connects to 10° synapses. Thus, there are in total 10'° synapses. One synapse can be represented in 4 bytes. Consequently, we need 4 million GBytes. Adding 1 GByte auxiliary storage yields 5 x 10'° bytes.

*We remember that our first computer programs were coded in hexadecimal code and later in Assembler.

systems. Nearly unlimited storage and computing power became accessible from any point in the world via cloud computing. Web pages were presented as hypertext documents using the hypertext markup language (HTML). These pages were addressed through uniform resource indicators (URI). Soon it became clear that each item in the real world might be addressable via URI’s—this was the birth of the Internet of Things (IoT). The IoT plays a major role in the development of digital twins, as we will see later in this book. And for digital selves, the implementation of avatars in the second life architecture in the year 2003 was a first practical step toward the virtualization of personalities in the web.

The connectivity of the web and the availability of nearly unlimited storage result in data oceans which cannot be controlled by humans exclusively. As a consequence, the buzzword “big data” represents a synonym for algorithms analyzing these data oceans. Similar to real life, “life” in a computer system means permanent sorting and searching.

It is well known that all the programming paradigms can be converted into each other. Thus, they serve as a means for the expression of algorithms. The above-described advances in computer science form the basis for solving one fundamental question which was posed by Alan Turing (1950) in the year 1950: “Can machines think?”. This question inspired numerous scientists, and the competition trying to solve the Turing test (Turing, 1950) had begun. This was the advent of artificial intelligence (AI).

Although Joseph Weizenbaum (1976) was convinced that a computer will never gain wisdom, he implemented one of the first AI programs, the ELIZA program. The rule-based program, implemented in the year 1966, used pattern matching to the written statements of patients to figure out how to reply. Although very simple, the program delivered astonishing results. Users had the feeling as if they were talking to a genuine psychotherapist. However, after a while, the program ended in loops because there were only a limited number of rules which were able to identify limited numbers of different words of the sentences entered via a console.

In the twentieth century, we implemented a version of ELIZA in our lab using the forward chaining production system language PAMELA (Barachini, 1991). The more rules we added to our knowledge base, the longer the user believed that he/she was talking to a psychotherapist. However, after a while, the game came to an end. Experts assert that today’s chat-bots do not better perform than early rule-based programs.

Cognitive architectures like GPS (1972) by Alan Newell and Herb Simon or SOAR from John Laird et al. (1987) and Laird (2012) used production systems* to implement means-ends analysis so that more general problems could be solved. The reinforced learning algorithm in SOAR, which dynamically modified the weight of rules, and the chunking mechanism enabling primitive learning could solve planning problems. A Truth Maintenance System guaranteed the consistent status of the working memory. The architecture is still used as a basis for cognitive simulations.

4 g 5 z : r Production systems consist of rules, a working memory (facts), and an inference engine.

For rule-based programs applied to a very limited scope of application, the computer pioneer Barr et al. (1981) coined the word expert system. The first expert systems were applied to medical problems before being widely used in industry. Expert systems delivered promising results, while cognitive learning had its limits.

In parallel to the above-described hardware and software developments, search algorithms have been developed. They are able to prune large search spaces. In combination with heuristics and high-performance supercomputers, they are able to solve complex problems. Some computer scientists believe that playing chess or Go is a good measurement for the effectiveness of artificial intelligence. In the year 1997, IBM’s deep blue computer has beaten the chess champion Kasparov. However, Go turned out to be a harder nut to crack. In this case, just speeding up the hardware and the implementation of search algorithms turned out not sufficient, since the search spaces are much more complex for Go than chess.

A new paradigm, artificial neural network processing, helped to solve the problem. Google’s AlphaGo Zero program used self-training reinforcement learning. AlphaGo Zero played itself in hundreds of millions of games such that it could measure positions more intuitively and collect its knowledge in neural networks. In a way, the network masters the game without teaching the rules. The so-called deep neural network was implemented on a single machine with four tensor processing units. In May 2017, the program could beat the world champion.

Many other applications have been developed with similar kinds of neural networks since then. Among them is AlphaFold which was able to predict the 3D structure of proteins in the year 2020. There are approximately 10°°° possible combinations for a single protein. It would therefore not be possible to solve this protein-folding problem with brute force algorithms in a reasonable time period, except for quantum computers which are in the experimental phase right now.

Similar good results were presented by IBM’s Watson supercomputer in 2011. This computer was doing better than any human when playing the jeopardy game. Encyclopedias, dictionaries, and thesauri were stored in local memory. The machine used different computing paradigms for problem-solving, among them neural networks. Numerous commercial and medical applications are currently on its way.

According to our observations, artificial neural networks were the last cornerstone for artificial learning. They go back to Frank Rosenblatt (1958) and to Hebb’s rule (Hebb, 1949). The rule explains human learning, in which simultaneous activation of cells leads to pronounced increases in synaptic strength between those cells. Much later, Teuvo Kohonen and others (Kohonen, 1982, 1984; Kohonen et al., 1991) presented forward and backward propagation algorithms for artificial neural networks. Having a closer look to the near future, autonomous cars, robots, and drones rely on this technology as well.

Whenever an article about a new Al-driven product is published, one can be confident that in most of the cases, neural networks are part of the game. However, the application of neural networks follows no cooking recipe. It is rather an experimental trial-and-error procedure until an appropriate network consisting of the right number of nodes, layers, and propagation functions is working properly.

Furthermore, engineers need to figure out how to present the data to the network and how to get the right sort of output that can be translated back into the real world.

Undoubtedly, deep learning with neural networks has brought new insights. AlphaGo Zero was able to produce strategies which were not known and applied by humans before. The same is true for radiological AI programs. They interpret pixels in different ways than humans. Therefore, we suspect that artificial intelligence deviates from human intelligence. However, if AlphaGo Zero would be confronted with a change of the board size during the game, the program would fail to deliver meaningful strategies. Thus, robustness is still a challenge for neural network applications. But there is no doubt that artificial intelligence can produce better and other results than human intelligence in limited domains.

Today’s implementations combine neural networks with rule-based, object-oriented, procedural, and functional approaches. However, the mix of the paradigms and the increasing architectural complexity make software vulnerable and errorprone. Gödel has shown that the set of Gödel numbers of functions whose domains belong to a class of recursively enumerable sets is either empty or contains all recursively enumerable sets. In other words, only trivial properties of programs are algorithmically decidable. Thus, the logical correctness of a program can be guaranteed only for very simple pieces of code. Therefore, we need to implement redundancy in our architectures, like it is provided by our natural brains.

When we come back on Turing’s question whether a machine can think, then the answer is ambivalent at the moment. Additional research in neuroscience is necessary. As long as neuroscience is not able to isolate the process that gives rise to consciousness in human brains, it will not be possible to create machines that have consciousness and mind, unless consciousness is a result of the thinking process. John Searle’s Chinese room experiment (Searle, 1980) clearly underpins this position. On the other hand, the Chinese room experiment encourages us to simulate understanding on computers in the sense of a weak AI. There is a difference whether the computer simulates a property or whether the machine understands what the simulation is about. In the latter case, we speak about strong AI. Eventually, new findings about the human brain will result from the human brain project which started in 2013 and ends in 2023.

The Turing test (Turing, 1950) requires that a machine must be able to execute all human behaviors, regardless of whether they are intelligent or not. That means that a machine that can think must include behavior related to emotions, cheating, lying, etc. Mind, understanding, and consciousness are elemental properties of thinking. In the second part of this book, we discuss if and to what extent digital selves might be able to adopt such natural properties.

From a long-term perspective on human civilization, i.e., a future time period in which human civilization could continue to exist, we refer to technological transformation trajectories. According to Baum et al. (2019), thereby “radical technological breakthroughs put human civilization on a fundamentally different course” (p.53). Some developments might persist as status quo, broadly similar to the current state of civilization, some cause significant harm (catastrophe trajectories), and some of them may lead to expansion of the human civilization toward other parts of the cosmos (astronomical trajectories). Which actions should be pursued depends on our existing portfolio and how reflective our practice develops toward systemic responsibility. In this book, we review fundamental parts of our current portfolio and reflect on capabilities that could actively shape our future.

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

The images or other third party material in this chapter are included in the chapter’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

Part I Digital Twins: Advent and Trans-human Development

This part lays a foundation to develop a socio-technical understanding of the nature of digital twins as representations and trans-human development objects.

Chapter 2 A) Background and Foundations cet

In this section, we provide some data on the origin of digital twins and provide insight why we consider them crucial for today when it comes to digital transformation. We also are projecting future developments in the context of humanoid socio-technical systems.

Grieves, who had raised the term in 2003, in his white paper “introduces the concept of a ‘Digital Twin’ as a virtual representation of what has been produced. Compare a Digital Twin to its engineering design to better understand what was produced versus what was designed, tightening the loop between design and execution” (Grieves, 2014). Given the background in industrial production, twin developers aim to increase transparency, and trigger efficiency gains, mostly in terms of faster process execution times, increased flexibility of organizing processes, and higher levels of security and safety.

As such, digital twins do not only designate a new technology but rather stand for the methodological integration of heterogeneous components, requiring protocols for exchanging even large amount of data between various technologies. Originally designed as virtual double of physical products, digital twins are increasingly becoming digital replicas of Cyber-Physical Systems (CPS). Not only the simulation of processes over the entire life cycle of a product but also envisioned architectures become core activities when CPS evolve (cf. Zhuang et al., 2017; Grieves, 2019).

A digital twin is consequently a virtual representation of a process, service, or product with parts of it in the real world. There are no predefined limits to what developers transform by digital means and represent in the virtual world: objects like planes including all kinds of operation characteristics, regions or urban infrastructures, or biological processes copying the human organism. Developers simulate functional behavior with the digital twin, in order to determine how capabilities can be changed depending on different parameters. This information helps to modify any product or service in a purely virtual manner before instantiating it for operation.