A study about the factors that contribute to the dehumanization of automated processes and the computer-human interaction.

(1)

A study about the factors that contribute to

the dehumanization of automated processes

and the computer-human interaction.

Imane Azrioual (11782358) - Imane_azr@live.nl Supervisor: Prof. dr. Tom van Engers

MSc Information Studies - Business Information Systems Faculty of Science - University of Amsterdam (UvA)

Submission date: 16-07-2018

Abstract. This paper aims to bridge the gap between the

dehumanizing effect of automated systems and the computer-human interaction. First, literature is reviewed, after which empirical research is conducted. The findings have been analyzed through the use of coding tables. Various factors that contribute to the dehumanization of automated systems and the computer-human interaction have been concluded from research and have been confirmed by the end-users. Trust seems to be the most significant factor from both theory and empirical research, since this factor is a concept on its own and has a relation with a great part of the other factors from the findings. It can be stated that automation can have dehumanizing effects on the end-users, and therefore have a negative influence on the computer-human interaction.

Keywords. Automation – Dehumanization –

(2)

Introduction

Technology has become a widely accepted phenomenon in today’s society. Internet of things, Block chain and Artificial Intelligence are a few examples of how far today’s technology has gone. For many years, movies have shown how robots have been created and how they have an active part in the daily lives of humans. Robots are being used more and more in various branches. There are autonomous cars that can trace danger from a great distance, drones that are used in different warfare fields and even Artificial Intelligence applications in medical healthcare.

Centuries ago, it was unimaginable to think that robots and computers could do or aide in the jobs of humans. However, in today’s society, AI and IT cannot be left out of our personal and work life. As stated by Fulk: ‘’The use of sophisticated communication technologies in the conduct of work is a common place in organizational life’’ (Fulk, 2008). By now most of us have experienced the effects of the rise of the information era. The effect of IT affects society, individuals and organizations (Steijn, 2004). Furthermore, IT has increased the productivity and reduced the need for technical experts. Centralization has become more popular, because of the flatter organizational hierarchies and the supervisory process becomes more formalized, with more reliance on measurable outputs and less on interpersonal processes.

Therefore, many people feel like just another number in their organization, because of automation, also called ‘’dehumanization’’(Steijn, 2014). According to Steijn (2014), computers reduce or eliminate the human element that was present in the pre-computerized systems. Automation and, therefore technology, has become an important strategy to reduce time and cost of various processes. However, the effect on other organizational factors needs to be researched more (Fulk. 1993). Haslam also explains that there are several types of dehumanization. However, technological dehumanization occurs because of IT. Also, he concludes that his model intends to extend the scope of dehumanization as a concept. Unfortunately, it is a general study on the concept dehumanization, while dehumanization in technological context is still not explained in detail.

Organizations in various sectors make use of IT to be as efficient and effective as possible. However, when it comes to processes that affect people’s lives in a certain way, it can have a different effect. Most organizations work with information systems and have automated multiple administrative and/or decision-making processes by using the systems. This leads to the question what relation dehumanization and automation have. Also, is it required to maintain a human factor in these processes to decrease certain errors? Furthermore, which factors do end-users consider as important and which ones as less important and does this match with the theory? How is the computer-human interaction between the end-users and the systems? And is it needed to maintain a human factor in these automated processes to fulfill the needs of the end-users?

(3)

The focus of this research will be put on the dehumanizing effects of automation and the computer-human interaction within health insurance organizations, because of various reasons. First, there is a scientific relevance behind this research. There is a lack of knowledge on the dehumanizing effects of technology and therefore, the computer-human interaction, which means that this research will help gain more knowledge on the computer-human interaction with the end-users of claiming systems and more knowledge on dehumanization. According to Ben-Ner and Urtasun (2013), more research is required to understand the effect of technology on skills, employees and the organizational processes (Ben-Ner & Urtasun, 2013). Appel et al. also explain that future research should involve a more detailed study on information visualization of the health insurance business (Appel et al., 2018).

Furthermore, Düll, König & Ohls state that insurers are equally sensitive to risks as banks are and even more sensitive than non-financial firms (Düll, König & Ohls, 2017). It can be concluded that health insurance organizations need to maintain a focus on risk points that can negatively affect the organizations, as the health insurance markets are vulnerable (Layton et al., 2017). Moreover, there is not much known about the defaults in the information systems of health insurance organizations, since most claims are handled automatically via a certain information system (Parasuraman, Molloy & Singh, 1993). In order to study and understand the relation between dehumanization and automation, it is important to identify the factors that contribute to the dehumanization of automated claiming systems and therefore, the computer-human interaction. The following research question has been formulated:

‘’What are the factors that contribute to the dehumanization of automated processes, according to the end-users of claiming systems of health insurance organizations?’’.

In addition to the scientific relevance of this research, there is a managerial relevance as well. It will benefit EY in the advisory of their clients on business process improvements from the perspective of the end-users (their clients), with the focus on automated processes.

This report consists of several chapters. The first chapter describes the theory behind this research. The second chapter explains the applied research methodology. After that, the results and analysis are presented, from which a discussion and conclusion follow. Finally, the advice, limitations and suggestions for further research are shared.

1. Literature review

Various theories have been researched thoroughly, in order to gain more insight in the main topics of this research, namely; Automation, Dehumanization, and Computer-human relation. First, the empirical state of knowledge is described. In addition, automation and dehumanization are described in paragraph 1.2.1 and 1.2.2 to understand the relation between the two topics. The last topic, computer-human

(4)

relation, is explained in paragraph 1.2.3. Finally, all theories are presented in a conceptual model.

1.1 Empirical state of knowledge

The effect of IT affects society, individuals and organizations. (Steijn, 2004). Furthermore, IT has increased the productivity and reduced the need for technical experts. Centralization has become more popular, because of the flatter organizational hierarchies. Also, the supervisory process becomes more formalized, with more reliance on measurable outputs and less on interpersonal processes (Steijn, 2014). According to Steijn (2014), computers reduce or eliminate the human element that was present in the pre-computerized systems. Montague and Matson (1983) state that this phenomenon involves the robotic pursuit of an approach to life and work that is unemotional, and therefore, ‘’unhuman’’.

1.2 Theoretical framework

In order to gain more insight in the computer-human relation, the relation between automation and dehumanization will be researched and explained first.

1.2.1. Automation

Lee & See (2004) have defined Automation as ‘technology that actively selects data, transforms information, makes decisions, or controls processes.’ Such technology has great potential to extend human performance, however, the systems cannot be trusted fully (Lee & See, 2004). Parasuraman and Riley (1997) describe automation as ‘computerized systems that perform activities that were previously performed by humans’ (Parasuraman & Riley 1997; Lyons & Stokes, 2012).

One of the main values of automation is to execute repetitive and complex tasks without errors. Automated processes can be exceptionally efficient, since people can focus their attention where it is needed. Risky environments also rely on automated process in place of people (Hoff & Bashir, 2015).

Parasuraman et al. (2000) explain that there are four primary automation types in a computer-human system (Parasuraman et al., 2000):

 Information acquisition: the automation of information acquisition relates to the identification and registration of input data. These activities fit in the first automation type, which is supporting the human process.

 Information analysis: The next type is automation of information analysis, and it involves cognitive functions. This second type is on a low level of automation where algorithms can be used to incoming data for extrapolation or predictions.

 Decision selection: The third type consists of decision and action selection and involves the selection of a decision among various selection alternatives. This type of automation encompasses varying levels of replacement of human selection of decision making by machine decision making.

 Action implementation: The final type, action implementation, refers to the actual execution of the action selection/ decision selection. This type of automation involves several levels of machine execution of the chosen action

(5)

implementation. Typically, this type is usually when a machine replaces a part of the human.

Steijn (2004) explains that computers reduce or eliminate the human element that was present in the pre-computerized systems. Automation has become an important strategy to reduce time and cost, however, there are other consequences that come with it (Steijn, 2004). The next paragraph will be focused on the reducing and elimination of the human element as a result of automation, also called ‘’Technological Dehumanization’’.

1.2.2. Dehumanization

‘Dehumanization is an important phenomenon in interpersonal as well as intergroup contexts, it occurs in and outside the domains of violence and conflict, and has social-cognitive dimensions in addition to the motivational determinants that are usually emphasized’ (Haslam, 2006). According to Burke, Bergman and Asimov (1985), technology provided by science to affect nature has also directly affected society (Burke, Bergman and Asimov, 1985). Montague and Matson (1983) also call this phenomenon ‘’a cultural condition of postmodern society’’, which involves the robotic pursuit of an approach to life and work that is unemotional, lacking in spontaneity (Montague and Matson, 1983). On the other hand, this robotic pursuit has a goal in terms of efficiency, conformity and automation-like rigidity. Critics explain that computers lack human emotionality, unpredictability and flexibility (Burke, Bergman and Asimov, 1985). Nissenbaum and Walker (1998) elaborate that the use of computers, will reduce social relatedness and increase standardization (Nissenbaum and Walker, 1998). In addition, humans have a need for empathy and trust and are usually able to judge the intensity of a human’s expression. However, this is not possible when certain processes are automated, in which there is no face-to-face contact (Etcoff & Magee, 1992; Hess, Blairy, & Kleck, 1997). Figure 1 show Haslam’s (2006) model, in which characteristics of ‘’normal’’ human nature are compared to characteristics of dehumanization due to technology.

(6)

Figure 1: Characteristics of mechanistic dehumanization. Source: (Haslam, 2006)

Automation can also cause people to feel less needed and eventually feel like automation takes their jobs, which can affect their mentality and therefore, their motivation and personality (Steijn, 2014). This theory will be kept in mind to test whether automation dehumanizes peoples work or as customers. Therefore, the self-determination theory (SDT) perspective is overviewed on automation and its influence on the computer-human interaction. This theory is taken as a central theory to explain how humanization is achieved and how dehumanization can be caused. If any of these needs is not taken into account, it is likely that one can feel dehumanized.

Self-Determination Theory

The self-determination theory (SDT) is a theory that concerns the human motivation and personality. This theory includes the motivation behind the choices people make, without any interference or influence. The focus is to understand the degree to which an individual is self-motivated and self-determined (Ryan, R.M. & Deci, E.L., 1985).

SDT is placed on the belief that humans have a natural positive attitude towards effort, commitment and agency in their lives. The theory calls this ‘’inherent growth tendencies’’ and identifies three needs that can lead to optimal growth and function. These three needs are: Autonomy, Competence and Relatedness.

 Autonomy: is the need to be in charge of one’s own life and to be in harmony

with one’s self employee Ryan, R.M. & Deci, E.L., 2000.

 Competence: is the need to be competent in what one does, to have the feeling

that one has control over a certain process Ryan, R.M. & Deci, E.L., 2000.

 Relatedness: is the need to be connected to others, to experience empathy and

the will to interact Ryan, R.M. & Deci, E.L., 2000.

The SDT divides motivation in two main domains: intrinsic and extrinsic. Intrinsic motivation is the inherent and natural (human nature) motivation to seek out new challenges and opportunities. Extrinsic motivation focuses on motivation that comes from external sources. The SDT consists of five mini-theories. Since automation is an external influence on the end-users in this research, one of these mini-theories will be described and explained.

(7)

Figure 2: Self-Determination Theory (SDT) model based upon the work of (Ryan, R.M. & Deci, E.L., 1985) From the theory, it has been shown that automation has become a robotic pursuit of an approach to life and work that is unemotional. Automation can cause a dehumanizing feeling to end-users. The next paragraph will be focused on these effects on the computer-human interaction.

1.2.3 Computer-human Relation

Dirks and Ferrin (2001) state that management and psychological literatures have long discussed the nature in which people rely on others (Dirks & Ferrin, 2001). The concept of trust is fundamental to the perception of reliance between people (Mayer et al., 1995). Similar to the human-human reliance research, trust is also essential in human-machine reliance (Muir, 1994). Thus, it can be stated that one’s trust in an automated system influences the reliance on that said system (Lyons & Stokes, (2012). Muir (1994) studied trust in automation, with the focus on a few factors, namely competence, responsibility, predictability and dependability (Muir, 1994). These factors represent the functionality of an automated system. In addition, other research focused on factors that influence trust in an automated system. These factors are:

 automation reliability (Rovira, McGarry & Parasuraman, 2007)

 system transparency (Dzindolet et al., 2003)

 workload (Biros, Daly & Gunsch, 2004)

 self-confidence (Lee & Moray, 1994),

 personality (Merritt & Ilgen, 2008)

 error feedback (Bisantz & Seong, 2001).

Little research has been conducted to distinguish the factors behind the overlap between human-human reliance and human-machine reliance (Lyons & Stokes, (2012). Madhavan and Wiegmann (2007) concluded that there are different mechanisms that drive the human-machine reliance (Madhavan and Wiegmann, 2007). Dix (2005) explains that when a system is difficult, the users’ motivation and productivity could drop. In that case the user will reject the system. Also, the interface seems to be a significant factor, since it can be seen as the mediator between the user and the system.

Self-determination

Theory

(8)

Dix (2005) also states that when one creates a system, it is important to always concentrate on the user (Dix, 2005).

1.3 Conclusion literature review

As can be concluded from the theory, humans have certain social needs, such as empathy and trust. However, it is not feasible for a computer to have these kinds of emotions. These are two of the factors that can lead to dehumanization and eventually a negative computer-human interaction.

However, it is not clear which factors contribute to this and whether empathy and trust are factors as well.

1.4 Conceptual framework

A few factors have been identified from the theory. According to the literature, it is clear that automation can cause a dehumanizing feeling to end-users, when certain human characteristics are not supported well. In addition, as can be seen in paragraph 1.2.3, the Computer-human relation, the connection between automation and dehumanization knows several factors.

The concepts and factors from the theory that will serve as the basis for the empirical part of this research can be found in the conceptual model in figure 3.

Concept Aim Factors Source

Automation Determine knowledge on automation and the automated system of health insurer.  Information Acquisition  Information Analysis  Decision selection  Action Selection (Parasuraman et al., 2000) Dehumanization Determine knowledge on dehumanization and explore the degree to which this applies to the automated process.  Technological dehumanization (Ryan, R.M. & Deci, E.L., 1985; Haslam, 2006) Computer-human relation Gain insight in the perception of the end-users about the system.  Trust  Responsibility  Predictability  Dependability (Muir, 1994)

Study the factors that can influence the  Automation reliability  System (Rovira, McGarry & Parasuraman, 2007; Dzindolet

(9)

Exploring databases and the Gap

Desk- Research

Empirical state of knowledge, theoretical framework and conceptual model

Literature

Review

Based on the factors in the conceptual model

Semi-structured

Interviews

Test factors of theory & individual interviews

Group Interview

Credibility Transferability Dependability Confirmability Ontological Authenticity

Trustworthiness

& Authenticity

trust of the end-users in the automated systems. transparency  Workload  Self-confidence  Personality  Error feedback et al., 2003; Biros, Daly & Gunsch, 2004; Lee & Moray, 1994; Merritt & Ilgen, 2008;

Bisantz &

Seong, 2001) Figure 3: Conceptual model

2. Research process and methodology

The applied research methods will be discussed in this chapter. First, a description of the context of this research will be given. In addition, information about the research design, analysis and sampling design will be shared.

2.1 Context of the research

This study is carried out within a period of three months, from May until July 2018 in Amsterdam, the Netherlands. This research is carried out on behalf of University of Amsterdam (UvA) and EY Cyber Security.

2.2. Research strategy

This research is conducted in a qualitative approach. The goal is to gain deeper insight in the computer-human interaction in automated processes. This study needs an approach in which theory guides research. Therefore, the deductive approach is used (Bryman, 2012). The online database of UvA and Google Scholar have been used to gather relevant theory. The following queries have been used: ‘’Automation’’, ‘’Dehumanization’’ and ‘’Computer-Human Interaction’’.

In addition, empirical research is conducted, in order to study the research question from a practical point of view.

Figure 4: Methodology Process

(10)

First, research is done in order to build the literature review, which consists of the empirical state of knowledge, a theoretical framework and a conceptual framework. In order to gain more insight in the computer-human interaction in the automated claiming process of health insurance organizations, semi-structured (expert) interviews with experts are conducted. In addition, a group interview is organized to gain additional information and to test factors from the results of the expert interviews. The methodology is illustrated in a roadmap in figure 4.

Firstly, new data is gathered through semi-structured interviews with experts in automation and computer-human interaction. Semi-structured interviews are used because of the possibility to ask further questions, since in depth information is needed. These interviews don’t have a completely pre-determined structure. Next to a few general questions, the interviewer has a list of concepts or topics that have to be covered at the least. This list is also called a guide or aide-mémoire.

The sequencing and phrasing of questions varies from interview to interview (Bryman, 2012). In addition to the list of topics also, the research question will be discussed.

Furthermore, another goal of the interviews is to gain more insight in dehumanization of automated processes. Finally, depending on the response, further questions will be asked. The outcome of the interviews are supplemented by the outcome of the group interview. The interviews aide more insight in the factors that have a role in the computer-human interaction, whereas the group interview focuses on factors that influence the end-users.

The interviews have been transcribed, after which they have been coded. The interviews have been analyzed by using selective and open coding. Open coding is used, because raw textual data is grouped into concepts, to research further and selective coding has been done because of pre-defined coding labels from theory (Figure 3) and as with time a few factors have been identified. These coding labels have been labeled as ‘’factors’’ and phrases from the interviews are grouped into these factors (Khandkar, S. H., 2009). The coding tables can be found in Appendix 1, 2 and 3.

2.3 Sample design

The general rule in qualitative research is that there is no standard sample size. One continues to sample until ‘’theoretical saturation’’ is reached (Hennink et al., 2016). Bryant and Charmaz (2007, p.611) describe theoretical saturation as ‘’the point at which gathering more data about a theoretical construct reveals no new properties nor yields any further theoretical insights about the emerging grounded theory’’(Bryant & Charmaz, 2007). In addition, Hennink et al. (2017) state that theoretical saturation ensures that all constructs of a phenomenon are fully explored and supported so that emerging theory is valid and robust (Hennink et al., 2017).

The semi-structured expert interviews were held with experts in the field of Automation and Computer-Human Interaction. The interviewees consist of professors from various universities in the Netherlands and professionals from well-known IT organizations. Also, these people have been appointed by their organization as the right person to discuss these questions. Furthermore, they have much experience in this field, in order to share insights in the experiences and changes of automation and computer-human interaction.

(11)

For the group interview, the group consists of six to twelve people who have experience with the online claiming systems of their health insurance. Everyone in the Netherlands is obligated to be insured from the age of eighteen, which means that the minimum accepted age for the group interviews is eighteen as well. In addition, the design of the interviews is a single-category design. This design focusses on the same category through all interviews, in order to gain new insights (Krueger & Casey, 2015). For the sampling, purposive sampling is used, since the goal is to sample participants whom are relevant to the research. Furthermore, snowball sampling is used to reach other relevant participants through the first couple of participants.

To ensure transparency, all interviews are recorded and transcribed (See Appendix 4), after which a coding table (See Appendix 1, 2 and 3) is created. Due to a limited timeframe, seven interviews and a group interview are held. However, in order to ensure quality, the interviewees were carefully selected.

Finally, the interviews are designed in such a way, that the interviewees are encouraged to have a comfortable conversation to share their knowledge and perspectives on several concepts and factors (Jones et al., 2010).

The interviewees and organizations they represent, have been anonymized due to privacy regulations and their request.

2.4 Trustworthiness and Authenticity

Lincoln and Guba ( 1985) and Guba and Lincoln (1994) proposed two primary criteria to assess qualitative research, trustworthiness and authenticity (Guba & Lincoln, 1985; 1994). One of the criteria of trustworthiness is credibility, which focuses on the credibility of the findings (Bryman, 2012). Credibility is ensured by respondent validation. The interviewees have received an overview of their answers and have confirmed it. Additionally, another criteria of trustworthiness is transferability. Transferability is the equivalence of external validity (generalizability) in quantitative research and could be an empirical issue (Lincoln and Guba, 1985). However, a thick description of the research is provided to assist in making a judgement about the transferability of these findings to other contexts (Bryman, 2012). In addition, dependability is the third criteria and is guaranteed by ensuring complete records of all phases of the research process. These records have been audited by a peer and can be found in Appendix 1, 2, 3 and 4. Finally, confirmability is the last criteria and is guaranteed by having an assistant during the interviews to guard the objectivity of the research questions (Lincoln and Guba, 1985).

In addition, the research is authentic, because this research fairly represents different viewpoints, as professors, professionals and end-users have been interviewed. Furthermore, it can be stated that ‘’ontological authenticity’’ is used, because this research helps different members to have a better understanding of their and other point of views, as different perspectives have been tested in this research (Bryman, 2012). 3. Analysis & Results

Interviews are conducted in order to gain more insight in the factors that contribute to the dehumanization of the automated health costs claiming systems of health insurance organizations. Questions are aimed to test theory and to gain more knowledge on other unknown factors. In addition, depending on the interviewee and situation, additional

(12)

questions are asked to gain a deeper understanding of the computer-human interaction. The questions are divided in three main concepts (Automation, Dehumanization and Computer-Human Interaction), which are defined in the theoretical framework. Furthermore, the outcome of the group interviews and the factors from the theory are discussed respectively in the following paragraphs.

Theoretical saturation is reached with three interviewees, after which four have been conducted to ensure saturation. In total seven interviewees, 5 professors in the field of automation information management and computer-human interaction (Professor #1, #2, #3, #4, #5), and two Information Technology professionals (GT #1 and GT #2) have been interviewed. Theoretical saturation can be confirmed, since no new concepts have arisen.

Also, a group interview is conducted to confirm the factors that are significant to the end-users. The interviews are anonymized, and the coding tables can be found in Appendix 1, 2 and 3.

3.1 Expert Interview

The individual interviews consist of experts in the field of information, automation and computer-human interaction. Their roles vary from professors and researchers at universities to professionals who work at well-known IT advisory organizations. The following paragraphs share the findings within every main concept of this research: automation, dehumanization and computer-human interaction.

3.1.1 Automation

In the interviews the distinction between various factors is acknowledged for the concept ‘’automation’’. Six factors that support the use of automated processes are distinguished. Firstly, the factor ‘’effectiveness’’ is discussed by most interviewees. Professor #1 explains that automated systems are effective in many tasks, especially administrative processes. Furthermore, efficiency is another factor that plays an important role in the use of automated systems. Professor #3 states that: ‘’Waarschijnlijk zijn ze, als ze goed ontwikkeld zijn, zijn ze minder foutgevoelig, ze zijn vaak sneller, vanuit de kosten zal het ook goedkoper zijn, de meer traditionele effecten1_{’’ (Interview Professor #3). The default chances and costs are lower and}

machines are faster in comparison to the use of humans. In addition, Interviewee GT #1 explains that ‘’ Automation influences organizations in such a way that certain work processes could be optimized, making the organization more profitable in the long run’’ (Interview GT #1).

Another main factor, which is discussed by four of seven interviewees, is the factor ‘’cost’’. Automation is supported by the belief that automated processes lead to a decreasing need for employees, thus a decrease in costs (Interview GT #1). However, 1 ‘’Presumably, they are, if they are well-developed, less error-prone, often faster and probably cheaper,

(13)

the interviewees also state that a decrease in costs doesn’t automatically mean a decrease in health insurance costs for the clients.

In addition, a fourth factor is identified, that is ‘’Distance’’. Professor #1 explains that an organization can use automated systems to alienate the clients from the organization, resembling a wall. However, this statement is not supported by other interviewees.

The factor ‘’flexibility’’ is the fifth factor that is mentioned often during the interviews. Flexibility involves the level in which a system is flexible enough to handle various cases, from standard cases to unique cases. Professor #3 explains that in administrative processes, less flexibility is needed. However, when dealing with people, such as in the health sector, it is important for a system to be able to differentiate between a standard case and a unique case.

Professor #4 states ‘’ De meeste gevallen in dit soort processen, zijn gevallen die je eerder hebt gezien. Maar er zijn altijd uitzonderingen. Dus wat je moet doen, is zorgen dat die standaard gevallen er soepel doorheen kunnen, maar ook dat een systeem slim genoeg is om te herkennen dat het een niet-standaard geval heeft… Dat lijkt me een basis uitgangsprincipe te zijn bij welke mens-machine interactie dan ook2_{’’(Interview Professor #4).}

Finally, functionality is discussed as well. Functionality involves the functionality of the system as well as the knowledge of the end-users on the functionality of the system. GT #1 explains that a negative computer-human relation can be caused by the lack of knowledge of the user with regards to the back end of the process or the system (Interview GT #1).

Thus, it can be stated that automation is depended on a variety of factors, to be successfully used and perceived by the end-users. However, automated systems can be perceived as dehumanizing, when certain factors are lacking or poorly executed. The next paragraph will describe and analyze the factors that contribute to the dehumanizing effect of automated systems.

3.1.2 Dehumanization

The interviewees made a distinction between four factors within ‘’dehumanization’’. However, two of the four factors have been mentioned several times and confirmed. The first factor is ‘’Transparency’’. Transparency is the level in which a process or the reasoning behind a decision is open to the end-user. Five of seven interviewees shared multiple insights in the factor Transparency. Transparency can have a dehumanization effect on the end-users. GT #1 explains: ‘’This also shows that even though people usually don’t know the details of a certain process, they still feel like they have to accept the outcome3_{’’ (Interview GT #1). A low level of transparency can cause the}

2 ‘’Most cases in these kinds of processes, are situations that have been witnessed before. However, there are always new types of cases. So what you need to do, is make sure those new cases can go through the system easily. But also that a system is smart enough to recognize a new type of case… This seems like a basic principle for any computer-human interaction’’.

(14)

end-user to feel a loss of control, which is the second factor. Control is the factor that decides whether the end-user has a certain level of control on the outcome of the process.

Professor #4 also explains that control is one of the most important factors to humans (Interview Professor #4). Furthermore, Professor #3 states that health insurance costs sometimes are declared automatically which leads to either a compensation or a collection. The end-user has no control over this process and often receives a short message afterwards (Interview Professor #3).

Finally, Professor #1 adds another factor, which is the Decision-making level of an automated system. The explanation behind this factor is that the level of dehumanization is also caused by the level in which a process or task is automated. An administrative process has a low dehumanizing effect on the end-users, but a health costs claiming process can be more dehumanizing (Interview Professor #1).

An automated system can be dehumanizing, according to the level of automation. However, what does automation mean to the computer-human interaction? The next paragraph will be dedicated to the factors that contribute to the computer-human interaction as an extension of the decomputer-humanizing automated systems.

3.1.3 Computer-Human Interaction

Eight factors that contribute to the computer-human interaction are identified. The first factor is ‘’Feedback Functionalities’’. This factor is mentioned by five out of seven interviewees. Professor #3 explains: ‘’ Als ik nadere informatie wil hebben dan kost het dus veel moeite om dat boven de tafel te krijgen, dan zit je zo een half uur aan de telefoon4_{’’ (Professor #3). In addition, Professor #1 explains that the Computer-Human}

Interaction can be negatively influenced by the lack of explanation when rejecting a claim.

Furthermore, a factor that is mentioned a few times is ‘’Technology Adoption’’. Professor #1 and Professor #3 state that when an end-user has a natural dislike for technology, then the Computer-Human Interaction is negatively influenced. It is also indicated that the Computer-Human Interaction with elders can be difficult to build, because of the frequent introduction of new technologies, which they often do not follow. This can also lead to the exclusion of certain groups in society.

Additionally, the factor ‘’Communication’’ is mentioned the most during the interviews. Communication involves the choice end-users have to communicate with the organization. Firstly, the communication language is raised, as it is focused on the higher-educated part of society (Professor #3). Furthermore, there is too little communication focused on how certain processes are executed, why a decision is made and what the end-users can do to avoid similar issues in the future. Also, another part of society can have a hard time understanding the systems, because of their age or 4‘’If I want more information, I would have to make a lot of effort before I get it, you would be on the phone for half an hour easily’’.

(15)

education level (Professor #1). Moreover, automation causes many counters and offices to close. However, personal contact is still valued.

Another factor is ‘’Limitations/Freedom of choice, which is mentioned by two interviewees. Professor #1 states: ‘’We moeten het nu maar anders doen en dan doen we het op een mobiele telefoon''. Waarom kan dat niet op een laptop? Nee dat kan niet5_{’’ (Professor #1).}

The following factors are ‘’Privacy’’ and ‘’Trust’’. These factors are mentioned several times as well and according to several interviewees, these factors have a causal relation with each other. Professor #2 explains that when ones privacy is harmed, then trust is also negatively affected (Interview Professor #2).

In addition, it is stated that the automated systems need to have a good working explanation function, as not everyone can work easily with a digital system. Professor #3 discusses that one of the design criteria should be a kind of ‘’emergency button’’ for people who don’t know what to do or don’t know what certain terms mean, so that they can be assisted (Interview Professor #3).

Moreover, the factor explanation functionality is also shared by the interviewees and is concerned with the explanation of the systems. Professor #3 discusses: ‘’ En ik denk dat een van de design criteria moeten zijn dat je een noodknop hebt op zo'n ding, dan moet je dus niet aan je lot overgelaten worden, een mogelijkheid om hulp te krijgen. Dan moet er even iemand mee kijken6_{’’ (Interview Professor #3).}

Finally, empathy is one the most mentioned factors and is concerned with the ability to understand and share feelings of another. GT #1 explains that technology doesn’t have any empathy (Interview GT #1). Professor #4 states that empathy and control are very important to humans and if these two lack, then they should change their health insurer (Interview Professor #4). Additionally, GT #2 states: ‘’ Technology will always have the need for people to develop, guide, install, manage and perform maintenance. Therefore, I believe that certain sectors in which organizations deal with people in a different way than employees or customers, computers will need a human factor to support the system and to still have the empathy factor integrated in the process’’ (Interview GT #2).

3.2 Group Interview

The group interview was held with end-users of different health insurance organizations and provided various insights in factors that play a role in the dehumanizing effect of automated systems. Firstly, communication is discussed multiple times. The end-users explain that little information is shared regarding claimed costs. Rather than an explanation behind the decisions of the insurance companies, they receive short messages in which simply the outcome is shared. 5 ‘’We need to do it differently, so we will do it on a mobile phone. Why can’t it be on a laptop? No that’s not possible.’’

6 ‘’And I think that it should be a design criteria to have an emergency button in a system, so that the user can receive help’’.

(16)

Interviewee #2 states: ‘’ Het zou verder fijn zijn als mijn zorgverzekeraar meer met feedback zou werken, dus niet alleen een kort berichtje over of ik wel of niet bij moet betalen7_{’’ (Group Interview, Interviewee #2). Furthermore, control is shared as another}

factor that plays an important role in this research. The end-users believe that they don’t have any control on the process and they pay everything they have to pay, without checking. The reason behind this is that they have little knowledge on the details of the process. Therefore, transparency seems to be a dehumanizing factor as well. Interviewee #6 explains: ‘’ Ik geloof het gehele systeem niet. Het proces is niet helder genoeg voor mij, ze zijn totaal niet transparant.

Wie weet wat ze afspreken met ziekenhuizen8_{’’ (Group Interview, Interviewee #6).}

Interviewee #5 also states that if the transparency and communication level would be higher, his trust in the organizations would be higher as well (Group Interview, Interviewee #5).

Additionally, trust is a factor that is recurring in many answers of the interviewees. Interviewee #2 discusses that his trust is very low, which seems to have an effect on the computer-human interaction as well, since he doesn’t want to contact his insurer personally (Group Interview, Interviewee #2).

Furthermore, the system itself needs to be functioning well. The interviewees share that the system needs to have certain functionalities, which would ease the use of the system. Also, it would solve certain issues that cause a dehumanizing feeling for the end-users, such as transparency and communication. Interviewee #4 states: ‘’ Het systeem zou eenvoudiger en makkelijker moeten zijn, want velen in mijn omgeving begrijpen niet wat ze allemaal kunnen met die apps. Laaggeletterden en ouderen gaan dit allemaal echt niet begrijpen. Een gemiddelde burger begrijpt dit ook niet!9_{’’ (Group}

Interview, Interviewee #4). In addition, a flexible system can lead to less dehumanization for the end-users, as the system consists of strict rules, where the computer decides whether costs are covered by the insurer or not. The system is not flexible enough for unique situations, although not every case fits into the ‘’standard’’ processes of health insurance organizations. The interviewees state that these organizations need to work on their flexibility and empathy, as they work with patients who need understanding for their situations. Interviewee #4 explains: ‘’ Kijk, ik geloof dat IT een stuk beter is geworden, dat zie je ook in bijvoorbeeld supermarkten terug. Maar ik verwacht wel dat wij als patiënten gewoon goed behandeld worden, dat ze zich kunnen inleven in de verschillende situaties10_{’’ (Group Interview, Interviewee #4).}

7 ‘’It would be nice if my insurer would give more explanation, so not just a short message on whether I have to or don’t have to pay’’.

8 I don’t believe the system at all. The process is not transparent for me. Who know what they discuss with the hospitals’’.

9 The system should be easier and simpler, because many people in my environment don’t know what they can do with the apps. Low-literate people and older people will not understand this. An average citizen doesn’t even understand!’’.

10 ‘’Look, I believe that IT has become much better, you can see that in supermarkets, for example. But I expect that we as patients are treated good and that they have empathy in various cases’’.

(17)

Finally, privacy is chosen as another significant factor in the dehumanization and computer-human interaction of automated processes. The better the privacy, the higher the trust the end-users will have in an organization.

4. Discussion

The literature research ensured various factors that contribute to the dehumanization of automated processes, and therefore, the computer-human interaction. Haslam’s model (2006) displays characteristics of human nature, which shows similarities with the findings. A few of the characteristics are Emotional Responsiveness and Depth. These are confirmed with the factors Empathy and Feedback functionalities. The other factors of the model have not been confirmed by empirical research.

In addition, the self-determination theory has been confirmed, as autonomy, competence and relatedness have been mentioned several times, through the use of factors as Control and Empathy.

Two factors from the theory have been confirmed by almost all interviewees from the group interview and from the expert interviews. These factors are Trust and Transparency. Responsibility and reliability have not been clearly confirmed, since these two factors correspond more with the back-end of the systems. The end-users are less involved with the back-end of the system, as they particularly experience the front-end. Furthermore, the theory showed insight in the factors that influence trust in a certain system. However, most of these factors haven’t been confirmed by the interviewees.

Finally, the difference between the findings from theories and the findings from empirical research can be explained by the lack of theories in the field of the three main concepts and their relationship. The research findings add several factors to the theory, as they have been confirmed by multiple experts and end-users. An overview of these factors can be seen in the table in figure 4.

5. Conclusion

This research has been executed, in order to answer the following main research question ‘’What are the factors that contribute to the dehumanization of automated processes, according to the end-users of claiming systems of health insurance organizations?’’. Various factors have been identified that contribute to the dehumanization and computer-human interaction, both by (expert) interviewees and by the group interviewees. First, communication is described. The level in which end-users receive detailed explanation about their claims, is highly significant. Furthermore, transparency and control are mentioned various times, as more transparency in the process, would lead to a higher feeling of control. This causes a lower feeling of dehumanization and also relates to trust in a certain relation. Privacy is found as an important factor as well. The interviewees explain that guarding ones privacy, leads to more trust in the system by the end-users. In addition, because the end-users are people in need of help during certain treatments, empathy is shared as an important factor.

Moreover, the system itself is also discussed. The findings showed that the functionality and flexibility of a system are very important. The system should do what it is expected to do, and it should have an option to be flexible in certain (unique) situations, where the standard procedure is not enough for the end-users. Also, the

(18)

human factor remains important to the automated process, since a computer needs regular input to function well. The table in figure 4 contains all factors from theory and the interviews with the number of interviewees that supported every factor.

Finally, this research provides EY and their clients, the health insurance organizations, a better insight in the factors that contribute to the computer-human interaction with their end-users. From an academic perspective, it contributed to the available literature by studying a relatively new domain and defining new factors.

Theory Factors Expert Interviews

Factors Group Interview Factors

Trust (7) Dependability (2) Responsibility (1) Predictability (2) Reliability (1) Transparency (7) Effectiveness (2) Efficiency (6) Costs (5) Distance (1) Flexibility (3) Functionality (2) Transparency (7) Control (5) Decision-making level (2) Feedback functionalities (5) Technology adoption (3) Communication (6) Privacy (2) Trust (7) Empathy (5) Communication (3) Control (2) Transparency (4) Trust (3) Functionality (4) Flexibility (4) Empathy (3) Privacy (2)

Figure 4. Overview factors that contribute to the dehumanization of automated process 6. Advice

Various factors have been identified, in order to encourage and improve the computer-human interaction and to lower the decomputer-humanizing effects of automated systems. First, the systems need to be functioning according to the expectations of the end-users and to be flexible to various cases. This will improve the computer-human interaction, between the system and the end-users. Furthermore, the organizational environment needs to be focused on the activities that support the system, such as personal interaction with the clients. Therefore, investments in soft skills and technological knowledge of the staff, will eventually create more trust from the end-users, as the staff will be able to interact well with the clients/end-users.

Finally, the system needs to be updated regularly, so that new types of cases can be recognized by the system. Therefore, it can be stated that a human factor will stay

(19)

significant in the process. The human factor should stay up to date on the newest cases and feedback of the end-users.

This will improve the computer-human interaction, which will stimulate technology and human to grow further together.

7. Limitations

One of the concepts in this research is dehumanization. Little research has been found on this concept, which made it challenging to find a useful guideline. In addition, significant factors have been identified, however the influence of these factors as well as the relationships between them remains yet to be studied or examined. Finally, due to a limited time frame and the limited number of allowed words, it was not possible to describe all findings in more detail.

8. Further Research

Further research is needed to explore a number of factors more thorough. One of the most significant factors from this research is Trust. This factor is complex and a concept on its own. There is already a lot of research on trust, however, not in the context of dehumanization and computer-human interaction. In addition, multiple other factors from this research have an influence on trust, but these relations haven’t been concluded. The relationship between the factors and the influence of them remains yet to be studied. Hence, further research is needed. In addition, this research is executed from the perspective of the end-users. Therefore, more insight from the perspective of the organization is needed, in order to approach the research question from a different viewpoint. Moreover, further research is needed in order to examine the automated systems of the health insurance organizations and to explore the computer-human interaction within these organization more in depth.

References

Appel, A. P., de Santana, V. F., Moyano, L. G., Ito, M., & Pinhanez, C. S. (2018). A Social Network Analysis Framework for Modeling Health Insurance Claims Data. arXiv preprint arXiv:1802.07116.

Ben-Ner, A., & Urtasun, A. (2013). Computerization and skill bifurcation: the role of task complexity in creating skill gains and losses. ILR Review, 66(1), 225-267. Biros, D. P., Daly, M., & Gunsch, G. (2004). The influence of task load and automation

trust on deception detection. Group Decision and Negotiation, 13(2), 173-189. Bisantz, A. M., & Seong, Y. (2001). Assessment of operator trust in and utilization of

automated decision-aids under different framing conditions. International Journal of Industrial Ergonomics, 28(2), 85-97.

Bryant, A., & Charmaz, K. (Eds.). (2007). The Sage handbook of grounded theory. Sage.

Bryman, A. (2012). Social research methods. Oxford: Oxford University Press.

Burke, J., Bergman, J., Asimov, I., Langley Research Center., United States., & College

(20)

of William and Mary. (1985). The impact of science on society. Washington, D.C: Scientific and Technical Information Branch, National Aeronautics and Space Administration. Dirks, K. T., & Ferrin, D. L. (2001). The role of trust in organizational settings. Organization science, 12(4), 450-467.

Dix, A. (2009). Human-computer interaction. In Encyclopedia of database systems (pp. 1327-1331). Springer, Boston, MA.

Düll, R., König, F., & Ohls, J. (2017). On the exposure of insurance companies to sovereign risk—Portfolio investments and market forces. Journal of Financial Stability, 31, 93-106.

Dzindolet, M. T., Peterson, S. A., Pomranky, R. A., Pierce, L. G., & Beck, H. P. (2003). The role of trust in automation reliance. International journal of human-computer studies, 58(6), 697-718.

Etcoff, N. L., & Magee, J. J. (1992). Categorical perception of facial expressions. Cognition, 44(3), 227-240.

Evans, J., & Jones, P. (2011). The walking interview: Methodology, mobility and place. Applied Geography, 31(2), 849-858.

Fulk, J. (1993). Social construction of communication technology. Academy of Management journal, 36(5), 921-950.

Guba, E. G., & Lincoln, Y. S. (1994). Competing paradigms in qualitative research. Handbook of qualitative research, 2(163-194), 105.

Haslam, N. (2006). Dehumanization: An integrative review. Personality and social psychology review, 10(3), 252-264.

Hennink, M. M., Kaiser, B. N., & Marconi, V. C. (2017). Code saturation versus meaning saturation: how many interviews are enough?. Qualitative health research, 27(4), 591-608.

Hess, U., Blairy, S., & Kleck, R. E. (1997). The intensity of emotional facial

expressions and decoding accuracy. Journal of Nonverbal Behavior, 21(4), 241-257.Khandkar, S. H. (2009). Open coding. University of Calgary, 23, 2009. Lee, J. D., & Moray, N. (1994). Trust, self-confidence, and operators' adaptation to

automation. International journal of human-computer studies, 40(1), 153-184. Krueger, R. A., & Casey, M. A. (2015). Participants in a focus group. Focus Groups A

Practical Guide for Applied Research, edited by: Krueger, RA and Casey, MA, Sage Publications, Inc, USA.

Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry (Vol. 75). Sage. Lyons, J. B., & Stokes, C. K. (2012). Human–human reliance in the context of

automation. Human factors, 54(1), 112-121.

Madhavan, P., & Wiegmann, D. A. (2007). Similarities and differences between human–human and human–automation trust: an integrative review. Theoretical Issues in Ergonomics Science, 8(4), 277-301.

Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational trust. Academy of management review, 20(3), 709-734. Merritt, S. M., & Ilgen, D. R. (2008). Not all trust is created equal: Dispositional and

history-based trust in human-automation interactions. Human Factors, 50(2), 194-210.

Montagu, Ashley, and Floyd Matson. (1983). The Dehumanization of Man. New York: McGraw-Hill. A cultural critique that articulates the psychological and social ills produced by modernity.

Muir, B. M. (1994). Trust in automation: Part I. Theoretical issues in the study of trust and human intervention in automated systems. Ergonomics, 37(11), 1905-1922.

(21)

Nissenbaum, H., & Walker, D. (1998). A grounded approach to social and ethical concerns about technology and education. Journal of Educational Computing Research, 19(4), 411-432.

Rovira, E., McGarry, K., & Parasuraman, R. (2007). Effects of imperfect automation on decision making in a simulated command and control task. Human Factors, 49(1), 76-87.

Parasuraman, R., Molloy, R., & Singh, I. L. (1993). Performance consequences of automation-induced'complacency'. The International Journal of Aviation Psychology, 3(1), 1-23.

Parasuraman, R., & Riley, V. (1997). Humans and automation: Use, misuse, disuse, abuse. Human factors, 39(2), 230-253.

Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). A model for types and levels of human interaction with automation. IEEE Transactions on systems, man, and cybernetics-Part A: Systems and Humans, 30(3), 286-297.

Ryan, R. M., Connell, J. P., & Deci, E. L. (1985). A motivational analysis of self-determination and self-regulation in education. Research on motivation in education: The classroom milieu, 2, 13-51.

Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American psychologist, 55(1), 68.

(22)

Appendix 1. Coding Table Theory

(23)

Appendix 2. Coding Table Expert Interviews

(24)

Appendix 3. Coding Table Group Interview

(25)

Appendix 4. Transcripts

(26)

Appendix 5. Interview Questions Group Interview Questions Female

Male

In what age- group are you?

- 18-26

- 27-40

- 40-55

- 55 +

When claiming your health costs, how do you do this?

- By mail

- Online

Health insurance organizations

- Which health insurance do you have?

- Why did you choose that one?

- What are the benefits of your health insurer?

- What are the disadvantages of your health insurer?

- Which factors are most important to you in the search of a new health insurer?

Automation

- Do you know what automation is?

o If yes, how did you learn about it?

- What do you think of automation?

- What are the pros and cons of automation?

Technical Dehumanization

- Do you know what (technical) dehumanization is?

o If yes, how did you learn about it?

- Do you think that automation and (technical) dehumanization have a relation?

Claiming process

- How do you claim your health costs?

- What is/are the benefit of that?

- What is/ are the disadvantages of that?

- Which factors are most important to you in the claiming process?

- Are these factors included in your claim handling?

(27)

- What do you think happens at the insurer when you hand in your claim?

- Have you ever experienced a mistake in the claim handling?

- What was the mistake and how was it fixed?

- Do you think that these types of mistakes could be prevented if a human factor

was added to the process?

- What do you think is the process behind sending a claim request?

- Which of these factors is most important, second, third and last?

o Trust

o Responsibility

o Predictability

o Dependability

- Which of these factors influence your trust in the system most?

o Automation reliability o System transparency o Workload o Self-confidence o Personality o Error feedback

(28)

Semi-structured Interview questions Automation