The Digital Twin of Humans

The necessary integration of human digital twins into the conception and description of digital working worlds in Industry 4.0 requires interdisciplinary research approaches, which are described in the chapters of this handbook. We introduce the topic and give a brief overview of this handbook, including short abstracts of the individual chapters.

Today, it is possible to collect and connect large amounts of digital data from various sources and life domains. This chapter examines the potential and the risks of this development from an interdisciplinary perspective. It defines the ‘global digital twin’ of a human being as the sum of all digitally stored information and predictive knowledge about a person. It points out that, compared to the digital twin of a machine, the human global digital twin is far more complex because it comprises the genetic code and the biographic code of a person. The genetic code contains not only a simple ‘construction plan’ but also hereditary information, in a form that is difficult to read. The biographic code contains all other information that can be assembled about a person, which is obtained via data from cameras, microphones, or other sensors, as well as general personal information. When the growing wealth of information concerning the genetic code and the biographical code is properly utilised, insights from biology and the behavioural sciences may be used to predict personal events such as health problems, job resignations, or even crimes. Because our own interests and those of private firms are partly in conflict over the use of this powerful knowledge, it is still unclear whether the global digital twins of humans will become a liberating or disciplining force for citizens. On the one hand, human beings are not machines: They are aware of their digital twin and therefore are able to influence it throughout their lives. Because of their free will, human beings are in general difficult to predict. Dystopias of full control over individual behaviour are therefore unlikely to materialise. On the other hand, private firms are beginning to take advantage of the available digital twins of humans by monopolising data access and by commercialising predictive knowledge. This is problematic because, unlike machines, human beings cannot only benefit from but also suffer due to their digital twins as they attempt to shape their own lives. We illustrate these issues with some examples and arrive at two conclusions: It is in the public interest for people to be granted more property rights over their personal global digital twins, and publicly funded research needs to become more interdisciplinary, much like private firms that have already begun to perform interdisciplinary research.

Digitalisation offers companies the potential to shape their processes and services along the entire value chain in a more efficient, flexible, and resource-saving way. However, there are many risks that companies need to consider in their digitisation efforts. These risks include not only the risks associated with technical challenges but also the risks concerning the organisation and the people working within it, such as, e.g., a lack of competencies or inappropriate processes. Even if new technologies are successfully implemented, their effective performance depends to a large extent on compatible processes, the relevant competencies of employees, and employees’ acceptance. Therefore, it is important to identify the risks associated with digital transformation at an early stage so that necessary measures for risk mitigation can be planned and implemented. In this way, the appropriate conditions for successful digitalisation within the company can be established. To meet these challenges, it is necessary to move from traditional risk management to holistic socio-technical risk management. This chapter shows how the classic risk management cycle can be enhanced for digitalisation projects against a socio-technical background. In addition, methods and approaches are explained that can help to support the risk management phases of identification and analysis, assessment, and treatment in a targeted manner.

Artificial intelligence in decision-making is a topic of increasing importance in research and society today. The concept of digital twins of humans (digital representation of humans), is also related to automated decision-making, as these digital twins could be used for decision-making. The increasing use of automated decision-making raises the question how employees perceive this form of decision-making. Therefore, research on justice and fairness perceptions is introduced in this chapter. The central part of this chapter is a literature review on justice and fairness perceptions in automated decision-making. The 43 included studies highlight the comparison of automated decision-making and human decision-making, as well as design factors and individual characteristics that influence justice and fairness perceptions. The results regarding justice and fairness perceptions are mixed. However, both decision-making authorities also offer unique strengths in terms of perceptions of justice: Automated decision-making is seen as less biased than human decision-making, while human decision-making is seen as open to peoples’ views and is associated with respectful treatment. The chapter concludes with a discussion of open questions and implications for the design of automated decision-making systems. Finally, we discuss the multiple ways that justice perceptions are related to the use of digital twins of humans.

In the recent past, negative relationships between actors in social networks have increasingly been evaluated. In addition to the consideration of the roles of individual actors, the focus of the investigations is also on the properties of the entire network. For this purpose, networks are modelled as signed graphs; this is a graph-theoretic approach that was introduced in the context of social psychological studies on structural balance in networks. Socio-technical networks in the work environment can be represented by signed graphs. We highlight some recent mathematical approaches that can be used for the in-depth analysis and design of socio-technical networks. Indeed, these tools not only allow us to understand the weak and strong points of a network but also show the value of specific relationships and actors. This analysis can then reveal how to improve the efficiency of the network with the minimum effort and, eventually, how to build teams for specific tasks.

Digital assistance systems have become ubiquitous in almost all areas of life, supporting us in the execution of tasks at work or in our everyday life. More and more of them are also able to adapt to the needs of the individual user and the environment. Although this development offers new possibilities and provides many benefits, it also poses challenges, as these kinds of systems not only have an impact on the performance of the user for a specific task but also affect psychological factors such as the user’s mental workload, self-efficacy, or satisfaction with the system or the task at hand. Moreover, research has shown that inappropriate levels of trust in or reliance on a digital assistance system (i.e., too high or too low) can lead to serious errors and even accidents. To provide effective and acceptable adaptive assistance, technical systems must first keep track of the state of the task, the environment, and the user and then take appropriate actions at the right time in order to improve the performance criteria of interest. The latter involves some form of interaction with the user and may include, e.g., informing, suggesting, intervening, or taking over control. This chapter provides an overview of different adaptive assistance systems, assistance types, and applied strategies, and it discusses their potential importance for the design of adaptive assistance systems. A categorization of different types and systems is presented; they are characterized according to the areas of use, the applied methods and technological implementations, the input data, and the possible effects on the user or the interaction. Concrete examples from our own research on adaptive assistance in monitoring tasks will be discussed.

Digital assistance has increasingly been implemented to cognitively support employees and enhance efficiency in different working environments. However, implementations of current assistance systems are often too rigid to provide personalised support, which can restrict employees’work autonomy. Therefore, in this chapter we discuss opportunities to enhance human work autonomy in collaboration with digital assistance systems by facilitating adaptivity within these systems.  First, we discuss the concept of autonomy from both a sociological and a technical perspective. Second, we develop a theoretical conception of the interaction between human and system autonomy. We identify three theoretical scenarios and provide examples of autonomy distributions in which increasing technical autonomy (a) restricts human autonomy, (b) does not affect human autonomy, or (c) enhances human autonomy. Based on this conceptualisation, we suggest that adaptive digital assistance promotes flexible distributions of autonomy levels between human beings and systems, which better addresses and enhances human work autonomy. Proceeding from research findings, our previous work, and theoretical assumptions, we present a guideline with concrete design recommendations for adaptive digital assistance systems that focus on increasing human work autonomy. In this context, we also discuss under what conditions the digital twin of a human can contribute to adaptivity to provide specific support for employees and better interaction opportunities with digital assistance.

In this chapter, an approach to assisting manual assembly processes individually is presented. The objective is to reduce process times by giving individual instructions to the assemblers. In manual assembly, worker assistance systems are used for worker guidance and the simultaneous quality assurance of the assembly steps in real time. The assistance system guides the workers step by step through the assembly process and simultaneously ensures that no errors occur. In practice, it has been shown that the use of assistance systems increases the quality of the assembly process but also increases process times. The approach shown in this chapter overcomes this loss in efficiency by assisting workers individually. The assistance system learns from the process executions of the employees through the digital twins of the employees. The assistance system adapts to the behaviour of the employees and highlights process steps in which mistakes might occur. Thereby, the number of interactions with the assistance system is reduced and more specific assistance is given.

Assembly is the final step of production and has to adapt to changing requirements. Produced parts are assembled into a product of higher complexity with defined functions within a determined time. Workers are the central actors in future cyber-physical assembly systems. They are crucial to the success of the entire system. There is a variety of methods and models for planning specific aspects of assembly systems. Examples include workstation design, assembly layout, and task assignment. In these approaches, individual characteristics and human factors are insufficiently considered. Within this chapter, an approach for the integration of human factors into cyber-physical assembly systems is proposed. This approach is an extension for planning methods and models that is meant to optimise the performance and cost of the assembly system.

In digitalised workplaces, a multitude of data is generated via computer-assisted work (CAW). These data enable comprehensive digital representations of employees, also referred to as the digital twins of humans. The possible consequences of CAW for the social integration of employees in the workplace have hardly been studied. With our contribution, we want to gain a detailed picture of how the design, implementation, and individual appropriation of CAW can shape working relationships, and we want to determine what role human digital twins can play in social integration in the workplace. We distinguish between CAW for information and communication, which directly shape relationships, and CAW that has the potential to influence relationships indirectly through other workplace characteristics, including control, routine work, autonomy, privacy, datafication, and transparency. The current literature shows that risks like isolation and bullying outweigh opportunities like extended cooperation possibilities. Furthermore, merging CAW into human digital twins could increasingly jeopardise social relationships. However, there appear to be design, implementation, and individual appropriation opportunities that could allow CAW to prevent threats to social integration in the workplace or even to foster positive relationships.

How digital technologies will affect skills and training needs is one of the conundrums of the evolving Industry 4.0. Important answers have been provided over the past 20 years by a new perspective: the task-based approach within labour economics. This approach interprets a job as a set of tasks, and it exploits employee survey data to measure and compare jobs in terms of skill needs. This chapter presents insights gained from the extant literature within this approach, including our own research, and identifies groups of employees whose job prospects and training opportunities are especially vulnerable. In particular, the evidence suggests that digital technologies often replace human employees who perform routine tasks. The remaining jobs will tend to include more sophisticated tasks such as directing and planning. Employees in jobs involving more routine tasks will therefore most likely need considerable retraining and upskilling. For various reasons, employers are unlikely to support this group of employees in terms of training and retraining. At the same time, routine jobs are often highly specific—their sets of tasks differ from those of most available jobs. Therefore, employees with such skill profiles are less likely to easily find alternative jobs. In other words, we identify a triple risk of digitalisation for some workers, namely those who conduct routine tasks, have a very specific skill profile that limits their ability to move to other jobs, and receive little training and retraining. Our approach and findings have important managerial and policy implications. Based on the framework we suggest, employers could identify and support employees with considerable risk by tracking their skills in a competence management system, i.e., by examining the digital twins of employees in terms of skills. If employers are unlikely to offer enough further training and retraining, an appropriate public policy response would be to reorganise initial vocational training—more than in the existing system, young employees need to be equipped with an even more general set of tasks, which would allow them to switch jobs within (and perhaps across) occupations.

External crowdworking (CW) is paid online work mediated by specialised crowdsourcing platforms. This chapter provides an introduction to various aspects of crowdworking with a focus on German-language platforms, based on the literature and our own results from the ‘Digital Future’ research programme. We define CW as an employment relationship and distinguish it from other forms of (non-)regular employment. Findings from a survey among crowdworkers show that crowdworkers are heterogeneous in terms of socio-demographic characteristics, and that the consequences of CW for health and work-life balance are ambivalent. Various platforms that broker complex tasks have developed a new type of rating system that commits workers to the platform. Based on crowdworkers’ past performance record, they achieve a particular status level, such as ‘five stars’, which indicates a worker’s reputation and determines the pay they can expect, as well as the tasks they can take on. Such rating-based compensation systems rely on a digital twin of each crowdworker that is stored by the platform. Today, such systems are platform-specific and proprietary, with a possible lock-in effect for employees. Public rating systems that cover multiple platforms are an alternative that would enable workers to transfer their reputation to other platforms. Overall, this chapter sheds light on an important but still under-researched form of flexible online work and illustrates that a novel form of the human digital twin is at the heart of platform management, with controversial implications for workers.