The relevance of artificial intelligence on the financial sector
- as perceived by senior managers
A qualitative evaluation of the Technology Acceptance Model
Preslava Petrova
10827129Master Thesis MSc in Business Administration | Digital Business University of Amsterdam | Faculty of Economics and Business Supervisor: Prof. Dr. Hans Oppelland
Second Supervisor: Prof. Dr. Hans Borgman Date: June 22, 2018
Statement of originality
This document is written by Preslava Petrova who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Artificial intelligence is a disruptive innovation that has a lot of potential for organizations in the finance sector. The technology has a wide application of uses within the industry, including automation, risk and portfolio management and decision making. However, academic literature and practice show that many organizations are not succeeding in adopting artificial intelligence, because it is not only the technological challenges, an organizational change is required too. The aim of this research is to describe specific organizational factors that influence the successful adoption of artificial intelligence in organizations among the financial sector. For the purpose of this research a qualitative evaluation of the Technology Acceptance Model is used by means of semi-structured interviews. System characteristics, organizational characteristics and usage behaviour are the foundation of the assessment of the acceptance level of artificial intelligence in this research. Organizations are provided with these capabilities in order to successfully adopt artificial intelligence. Next, the research analyses the effect of the acceptance level of artificial intelligence on the implementation process of organizational strategy. The research shows that artificial intelligence benefits the implementation of strategy only when it is accepted internally. In addition, the application of artificial intelligence has to be further extended into the strategy formulation process. This research provides a foundation for future research and provides managers with insights on how to be prosperous in the adoption of artificial intelligence.
Keywords: artificial intelligence; information systems; technology acceptance; disruptive innovation; organizational perspective; strategy implementation
Table of Contents
I. Introduction……….1
1.1. Research problem……….……….……….3
1.2. Research objectives……….……….5
1.3. Research method……….……….………….6
1.4. Structure of the research……….……….7
II. Literature review………..7
2.1. Value creation ………8
2.1.1. Value creation through strategy………9
2.1.2. Strategy implementation: creating value through resources…….11
2.1.3. Information technology (IT) as a resource………12
2.2. Artificial intelligence……… 12
2.2.1. Definition……… 14
2.2.2. Applications of artificial intelligence in finance………15
2.2.3. Acceptance of artificial intelligence………16
2.2.4. Technology Acceptance Model……… 18
2.3. Conceptual Model……… 18
2.3.1. System characteristics………19
2.3.2. Organizational characteristics………20
2.3.3. Usage behaviour………21
III. Research design.………..21
3.1. Research strategy………22
3.2.Data collection techniques………22
3.2.1. Semi-structured interviews………23
3.2.2. Purposeful sampling ………23
3.2.3. Limitations ………24
IV. Data collection……….. 24
4.1. Sample………26
4.1.1. Selection criteria………27
4.1.2. Sample description ………28
4.1.3. Data collection methods ………29
4.2. Data analysis ………30
4.2.1. Coding………30
4.2.2. Analysis………30
4.2.3. Credibility, transferability and dependability ……… V. Results and discussion ………32
VI. Conclusions………45
Bibliography……….. 62
Appendices………….69
Appendix 1: Interview protocol………69
Appendix 2: Transcribed interview example………74
Appendix 3: Overview of codes………78
I. Introduction 1.1 Research problem
The concept of “artificial intelligence” (AI) - an information system using computer systems to perform tasks that ordinarily require human comprehension - has been present for more than 60 years. But it was not until recently that AI appeared on the verge of revolutionizing industries as diverse as health care, law, journalism, aerospace, and manufacturing, with the potential to profoundly transform the way people live and work (Russell and Norvig, 2016). As a sequence, technology leaders across the world have begun to discuss how artificial intelligence can be used in order to improve speed, quality and functionality, and potentially increase top line revenue growth (Luger, 2005). A convergence of forces has propelled artificial intelligence into the business mainstream, as artificial intelligence is the main system applied to process Big Data (O’Leary, 2013). According to a recent report from the Boston Consulting Group (BCG), involving more than 3,000 business executives, managers and analysts in 21 industries in 112 countries, more than 75% of business executives anticipate that artificial intelligence has the potential to establish competitive advantage or create new lines of business for their enterprises1. Business leaders have high expectations of successfully implementing and improving their organization with AI in the very near future. The head of Artificial Intelligence at KPMG, Shamus Rae emphasised on the importance of AI in this age and time and put an emphasis on the effect of AI particularly on the financial sector:
“There’s never been so much data at our fingertips—and arguably there’s never been greater internal and external pressure to analyse that data to manage compliance and risk, and in this context, AI is opportunity managers cannot ignore, offering companies the ability to process vast quantities of data at lower cost.” 2
1
Source: https://www.bcg.com/d/press/6september2017-gap-between-ai-ambition-execution-169791 Date of last approach: June 21, 2018
In addition to compliance, financial institutions and vendors have been adopting AI to assess credit quality, to price and market insurance contracts, and to automatize customer interaction (Campbell-Verduyn, Goguen and Porter, 2017). AI can discover behavioural patterns in data, which leads to data-driven decision making, with the ability to combine data from different departments, obtain internal and external data sources and thereby produce unique results. Executives consider AI as a driver of new business opportunities and reduced costs, and has the potential for companies to create value, which is the ultimate goal of organizations (McAfee and Brynjolfsson, 2012).
Even though the potential value that artificial intelligence can create, reports show that merely one out of five companies has integrated artificial intelligence in some processes, and only one in twenty companies has extensively incorporated AI into its current processes and operations3. Organizations struggle to become successful adopters of artificial intelligence and to use analytics to their advantage (Ransbotham, Kiron and Prentice, 2016). In order to employ artificial intelligence and utilize the available data for business purposes, the existing data systems have to be adjusted. This task is time-consuming, expensive and a challenge for developers (Ransbotham, S., Kiron, D., & Prentice, 2016). However, considering the innovative and disruptive nature of artificial intelligence, the implications are bigger than technology alone (Legris, Ingham and Collerette, 2003). Scholars state that the biggest challenges in adopting new technologies are mainly managerial and cultural by nature (Shein, 1984). Many organizations experience difficulties in understanding, implementing and communicating new technologies internally, which as a result prevents the organization from a straightforward adoption process (Russel & Norvig, 2016). The concept of artificial intelligence is a disruptive innovation and requires a massive change in the organization (Rogers, 2010). In order for value to be created artificial intelligence needs to be accepted by the organization.
3 Source: https://www.bcg.com/d/press/6september2017-gap-between-ai-ambition-execution-169791
Artificial intelligence has a lot of potential for organizations in the finance industry, but reality shows that adopting artificial intelligence is difficult. In such a dynamic environment as the finance sector, organizations need certain capabilities in order to become successful in the adoption of artificial intelligence. The place of artificial intelligence in the strategic business model has not yet been acknowledged. In order for organizations to become successful in the adoption of artificial intelligence, it is important to fully understand the concept of artificial intelligence so that it can be determined what capabilities are required to successfully adopt the artificial intelligence technology in the organization. This is an area that has not received a great extent of attention and uncovers an interesting literature gap. This research focuses on the role of artificial intelligence in the business model of companies among the financial sector, in order to assess what capabilities are needed in order to apply artificial intelligence to its full potential so that it can benefit organizations without disturbance. This needs to be addressed before AI can be successfully incorporated into critical organizational processes.
1.2 Research objectives
In a recent research by Laeven, Levine and Michalopoulos (2015), the authors conclude that financial innovation is necessary for sustaining economic growth. In other words, in the course of time technological innovation and economic growth will stop unless financiers innovate. The financial sector, including banks, investment funds, insurance companies, financial services and financial technologies (FinTech), is highly dynamic, due to the constantly-changing market conditions (Beck and Demirguc-Kunt, 2006). In addition, disruption from artificial intelligence is visible in the industry, which makes the financial sector an interesting field for research. Companies have recognised the potential artificial intelligence has and have been applying it to manage risk and make investment decisions, yet many executives are not certain about what exactly to expect from AI and the ways this state-of-art technology fits into their business model. Pressured by time and competition, the time to identify an organisation’s strategy is now (Moore, 1996). This research examines the ways artificial
intelligence influences the implementation of organizational strategy and analyses the adoption process of artificial intelligence, with the main goal to find when and under which conditions artificial intelligence helps organizations achieve their strategic goals. In addition, this research analyses the acceptance of artificial intelligence within an organizational environment with the aim to find whether artificial intelligence only benefit an organization when it is accepted internally.
This research aims at both a managerial and research contribution. At the moment, the available literature does not provide knowledgeon the relationship between artificial intelligence and strategy execution. The primary objective of this research is to establish a conceptual framework with a focus on the organizational perspective, that will point organizations in the right direction to become successful in the adoption and thereby implementation of artificial intelligence.Frameworks are based on the identification of key concepts and are used to clarify and propose relationships among these concepts (Järvelin and Wilson, 2003). For the purpose of this research, the conceptual framework will derive from the Technology Acceptance Model (Davis, 1985), which assesses the acceptance level of information systems within organizations. Models help to stimulate research and the extension of knowledge by providing both direction and impetus (Donovan, Bransford and Pellegrino, 1999). This research aims at contributing to the literature by further expanding and developing the theory of Davis (1985) and applying it to a new innovation technology: artificial intelligence. Furthermore, unlike most other research that is confined to the industry-wide perspective, this research focuses solely on the financial sector.
The modified conceptual model provides knowledge on the criteria that influence the adoption of artificial intelligence. The conceptual model contributes to literature by determining concrete capabilities that are relevant for the adoption process of artificial intelligence. Furthermore, the conceptual framework explains the observations gathered through the data collection process. The main purpose of this conceptual framework is to make the research findings meaningful and generalizable for other organizations within the financial sector. On the other hand, the practical
objective of this research paper is to provide baseline information on the way companies implement artificial intelligence, the prospects for its growth, and the steps executives have to take in order to create and cultivate a successful strategy for their business. This helps organizations acknowledge the importance of the transition process of implementing artificial intelligence and handle this transition in which many organizations are failing. The main goal for organizations is to develop strategies that are relevant, flexible and more effective (Moore, 1996). This research will help to synchronize a company’s strategy around a common set of objectives, aiming at a faster and better product-innovation capabilities. Hence, this research aims to provide management proposals for artificial intelligence implementation, in order to transform strategic thinking into actions and reach the organizational objectives more expeditiously, which is important for sustaining competitive advantage in a sector as demanding and dynamic as the financial one.
1.3 Research method
The aim of this research is to investigate the relevance of artificial intelligence on the implementation of strategy within organizations in the financial sector. This is performed by analysing the applications of artificial intelligence and the acceptance levels of employees, which is described by the attitudes of people towards the technology. Qualitative research uses the opinions, attitudes and beliefs of people, which is relevant for the subject of this research in order to obtain an insight into the personal experiences of the respondents. As stated by sociologist Cameron (Toye, 2015, p.1): “Not everything that can be counted counts, and not everything that counts can be counted”. This is the reason why qualitative research is valuable, as it allows to gain a broader understanding of the topic. Furthermore, qualitative research offers the possibility to obtain a more in-depth and personal assessment of the research objectives by hearing experiences, opinions and beliefs of individuals.
The research design of this Master thesis is qualitative and is performed by means of semi-structured interviews with open ended questions. This research is parallel, which means that “the cases
are all happening and being studied concurrently” (Thomas, 2011, p.517). In addition, the approach is abductive, by validating the theoretical framework and using the interviews to examine the conceptual model and explore new concepts and experiences that can be incorporated in the research. In addition, this study will be exploratory in nature. Exploratory research tackles new problems on which little or no previous research has been done (Saunders, 2011). Exploratory research is a valuable way of finding out “what is happening; to seek new insights; to ask questions and to assess phenomena in a new light” (Robson, 2002, p. 59). This is a suitable approach for this research paper, as even though a substantial amount of research has been done on artificial intelligence (Liebowitz, 2001), relatively little is known about its link to the execution of organizational strategy. In addition, when compared to descriptive and explanatory research, these research designs are mainly based on “the richness of the rival propositions in theories related to topic of the study” (Eisenhardt, 1989, p 541). Due to the lack of such propositions regarding the acceptance of artificial intelligence within financial organizations, the exploratory design is most suitable for this research.
1.4 Structure of the thesis
This research is organised as follows. Firstly, the literature is discussed starting with an exploration of the importance of strategy in organizations. The literature review continues with an examination of the concept of artificial intelligence. It emphasizes the relevance of artificial intelligence for the financial sector and highlights that adopting artificial intelligence is a transformational process and asks for changes in the ways companies operate. Furthermore, the literature review focuses specifically on the adoption of artificial intelligence. In the end of the literature review the conceptual model of this research is presented including the model’s modifications accustomed to this research. Afterwards, the method section is presented, followed by the data collection section. Next, the results of the conducted interviews are presented and discussed in accordance with the available literature. Finally, this research
ends with the conclusion, summarizing the main findings, contributions, limitations and suggestions for future research.
II. Literature review
In general, organizations adopt artificial intelligence because it has the potential to create value. This value is created by making the correct decision based on the analysed data. The potential value is unlocked when organizations actually use the technology. In order for artificial intelligence to be used, however, it must be adopted within the organization. Given the disruptive nature of artificial intelligence and the dynamics of the financial sector, this is a challenging task. To begin with, value creation is explored, followed by an evaluation of strategy as a tool to create value. Afterwards, the concept of artificial intelligence is discussed and the relevance of acceptance in the technology adoption process. The chapter ends with the conceptual framework of this research.
2.1 Value creation
The main goal of organizations is to create value, which is usually expressed through achieving profit and surpassing the competition. Organizational success is when the market demand aligns with the offerings of the organization, therefore firms focus on the markets they compete in (Day, 1994). Empirical findings confirm that organizations that are able to meet the market demand and respond to changing conditions better than competitors, achieve a competitive advantage and through that higher profitability (Day, 1994). Organizations nowadays are eminently dynamic and operating globally, and the business environment has become highly competitive. Therefore, in order not to become obsolete, organizations have to comprise change and correspondingly alter their game plan, also acknowledged as “strategy” (Teece, 2010). However, there is little academic research on the capabilities needed for organizations to be able to successfully execute their strategies and be prosperous. In order to determine what are the capabilities an organization needs
to be successful in the adoption of artificial intelligence, first an understanding of how value is created through strategy is needed. This is presented in the following section.
2.1.1 Value creation through strategy
The concept of strategy as applied to business has begun to appear with great frequency in the 1960s and has received a lot of attention from scholars of a variety of areas such as industrial economics (Chandler, 1962; Porter, 1980, 1985), organizational theory (Hall and Saias, 1980; Miles and Snow, 1985; Mintzberg, 1988) and management and consultancy (Ansoff, 1965; Schendel and Hofer, 1979; Hendersson, 1992). The definitions vary from one author to another, but in essence the interpretations converge in the idea of strategy as “the pattern in the stream of decisions and activities, that characterizes the match an organization achieves with its environment and that is determinant for the attainment of its goals” (Hofer & Schendel, 1979, p. 25). The importance is on the arrangement of activities that has an effect on the attainment of the organizational goals in respect to the surrounding environment (Håkansson and Snehota, 1989).
Strategies help companies understand their own organization by identifying core capabilities and weaknesses. Furthermore, strategies are valuable for understanding the organizational external environment, as comprehending what is happening in the industry can ensure growth and long-term profit (Miles, Snow, Meyer and Coleman, 1978). In addition, strategies create a vision and direction for the organization as a whole. A common purpose, clear goals and a set of actions required to reach them assures that everyone is working towards the same outcome, organizational success. Therefore, strategy is fundamental to the sustainability and success of any organization, regardless of the field that the organization is competing in (Miles et al., 1978). Strategy management is viewed as the process of adjusting the consequence of activities performed by an organization reflecting the external environmental conditions in which the company operates. Therefore, managing strategy is in essence the administering of the process through which the pattern of activities to be performed by the organization is perceived (strategy formulation), and then taking the necessary actions to ensure that
these activities are performed (strategy implementation), (Gupta and Govindarajan, 1984). The need for improved implementation of strategy by means of information technology (IT) has been emphasised in both empirical (Lederer and Sethi, 1992; Earl, 1993) and practical research (Galliers, 1993), which is the reason why this research investigates the effect of artificial intelligence on strategy implementation.
2.1.2 Strategy implementation: creating value through resources
Research on business strategy has been involved predominantly in understanding what makes a business organization effective in its environment, and investigating the organizational processes required to strengthen this effectiveness. Neilson, Martin and Powers (2008) state that “a brilliant strategy, blockbuster product, or breakthrough technology can put you on the competitive map, but only solid execution can keep you there, because you have to be able to deliver on your intent” (p. 2). The authors refer to strategy implementation as the results of thousands of decisions made by employees on a daily basis, acting in accordance to the information they have and their own self-interest (Neilson, Martin and Powers, 2008).
The level of successful implementation of strategy is measured by the match an organization makes between the opportunities and risks created by the external environment and its internal skills and resources (Grant, 1999). It is often emphasized that due to the constantly changing environment, this has to be an ongoing process (Pearce, Robinson, Subramanian, 1997). Furthermore, Grant (1999) states the accumulation of resources is supposed to be the precondition for the survival of the organization and therefore, strategies have to be designed in a way to exploit the maximum effect of a company’s unique characteristics. The resources that an organisation has can allow it to outperform others in the same industry or product market, also addressed as “competitive advantage”. According to Chaharbaghi and Lynch (1999), if an organisation correctly manages to identify its existing resources, focuses on their exploitation and thereafter develops a new generation of resources, it would
be able to sustain its competitive advantage. This is in line with the resource-based view of value creation, which focuses on the firm-level assets and capabilities an organization owns that can create sustainable competitive advantage (Day, 1994). The resources and capabilities originate and develop over time and the combination of these unique resources have to be utilized by the management level in order to create competitive advantage (Barney, 1991). However, it is not only necessary to focus on the resources an organization owns, but also consider the markets they compete in. This is a shortcoming of the resource-based view, acknowledged by Day (1994). The author introduces dynamic capabilities, which support a market position that is difficult to match and valuable (Day, 1994). Teece (1997) states that these capabilities provide a unique combination of building, integrating and reconfiguring internal and external competence, in order to keep up with the constantly changing environment. Day proposes five steps to enhance the capabilities: “(1) the diagnosis of current capabilities, (2) anticipation of future needs for capabilities, (3) bottom-up redesign, (4) top-down direction and commitment, (5) creative use of information technology, and (6) continuous monitoring of progress” (Day, 1994, p. 37). In addition, Håkansson and Snehota (1989) argue that “the effectiveness of an organization and its potential for accumulating resources is assumed to be a function of matching the characteristics of the environment with the capabilities of the organization” (p. 3). A positive balance in the exchange of resources with the constraints of the environment is ensured by adapting to the environment. The idea of ‘‘fit’’ between the capabilities of the organization and the characteristics of the environment (in particular customers and competitors, referred to as the ‘‘market’’) is the central theme in the strategy management doctrine (Miles & Snow, 1984; Venkatraman and Camillus, 1984). Piercy (1998) further elaborates on the constraints that the environment an organization operates in can possess over the performance of strategy, which can be time, culture, person or strategy specific.
Researchers have concluded that desired levels of performance cannot be achieved unless organizations respond effectively to relevant environmental demands (Emery and Trist, 1965;
Lawrence and Lorsch, 1967; Thompson, 1967). Since environmental demands diverge across organizations, different firms may have to emphasize the development of different key functional areas (Hitt, Ireland and Stadter, 1982). This is a topic that has been analysed in the existing literature, however the main focus is solely on information technology and not centred specifically on the financial sector. When establishing the fundamental capabilities an organization needs in order for artificial intelligence to be adopted, the importance of both the firm-specific assets and the market environment play an equally important role.
2.1.3 Information technology as a resource
Information systems (IS) are organized systems representing “an assembly of components which ensure data processing for the purpose of obtaining information, throughout collection, manipulation, process, storing, organization and distribution activities, in order to assure decision support to accomplish organizational objectives” (Georgescu and Jeflea, 2015). Information systems are implemented to elevate unique business competencies, restructure industries and facilitate global competition (Chan, Huff, Barclay and Copeland, 1997). Baets (1992) argues that organizations with more participation by the IS department in business planning have greater benefits compared to firms with lower IS participation. According to the author, these benefits include improved communication between different departments and hierarchical layers and a clear definition, enhanced alteration and adaptation of current overall corporate strategy (Baets, 1992). As a result, senior executives, strategic planners and IT managers have been paying attention to opportunities for achieving competitive advantage through information technology (Bakos and Treacy, 1986). One of the reasons behind this are the unstable economic conditions, also relevant for the financial sector, which creates a challenging business environment (Rockart and Morton, 1984). In addition, technologies such as information systems are offering a great array of capabilities at low costs and as a consequence firms’ abilities to utilize the technology are enhancing (Bakos & Treacy, 1986). As a result, in recent years, information
systems have become a driver of business strategy and a key tool for the execution of business strategy (Rathnam, Johnsen, Wen, 2005). Implementing information systems into the organizational processes has become a continuous management activity and “an on-going management activity and is an integral part of the dynamic process of defining and monitoring corporate strategy” (Baets, p. 211, 1992).
Technological advancement has been reshaping the finance sector for the past three decades, ensuring effectiveness and efficiency, security of transactions, internal processes optimization, raising products and services’ complexity degree, which has led to an advancement of financial information systems (Georgescu & Jeflea, 2015). In the early 1990s the Financial Services Technology Consortium was initiated - a project aiming to facilitate technological cooperation efforts. The term “Financial technology”, or “FinTech” originated, which refers to the use of technology to deliver financial solutions (Arner, Barberis and Buckley, 2015). Nowadays, this term refers to a large and rapidly growing industry, representing $197 billion in investment as of 2014 (Lodge, Zhang and Jegher, 2015). The digitalisation of financial services is offering incrеаsеd payment transparency, fаster settlement periods and hаs еnаbled grеаter аudit cаpаcity and rеducеd sеcurity risk. In аddition, the аccеss to rеаl-time custоmеr dаta hаs fаcilitаtеd products that аrе suitеd to uniquе customеr nееds (Philippon, 2016). Considering the rapidly growing part FinTech has in the functioning of finance, a great managerial attention is warranted, which makes the importance of IT alignment with corporate strategy even more relevant and therefore reveals an interesting field for research (Arner, Barberis & Buckley, 2015).
2.2 Artificial Intelligence 2.2.1 Definition
A compelling advancement in IT in the past decades is artificial intelligence (AI). AI is the technology and science established on assorted functions to develop a system that has the ability to work and think like a human being. It can analyse, reason, solve problems, learn and conclude. Kurzweil (1990)
defines artificial intelligence as “the art of creating machines that perform functions that require intelligence when performed by people” (p. 13). In essence, AI is the study of complex information processing problems that have their roots in biological information processing aspects (Marr, 1977). AI is based on various functions that develop a system that can think and work like a human being. It combines disciplines such as computer science, psychology, biology, engineering, mathematics and linguistics. AI can analyse, reason, learn, conclude and solve problems. The goal of AI is to develop computers that can stimulate the ability to think, see, hear, walk, talk and feel. In other words, simulation of computer functions normally associated with human intelligence, such as reasoning, learning and problem solving.
The managerial implication of AI is to obtain a knowledge-based computer system that will help managers to make more reasonable and quick decisions in business (AO'Brien, Marakas, Hills and Lalit, 2006). For instance, an investor might adopt an algorithm that differentiates cars to count the number of cars in a retail parking lot from a satellite image in order to predict a probabilistic store sales figure for a certain period. The computational power of deep learning algorithms has been noticed by major tech companies, who started acquiring deep learning start-ups and began investing heavily in deep learning research. In 2015, The World Economic Forum reported that global investment in AI start-ups rose from $282 million in 2011, to $2.4 billion (Figure 1). This exhibits the growing interest of organizations to integrate AI in their processes.
Figure 1: Global artificial intelligence merger and acquisition activity, 2012-20174
2.2.1 Application of artificial intelligence in finance
The growing use of FinTech has contributed to the adoption of AI and machine learning in financial services. Financial market participants have benefitted from faster processor speeds, lower hardware costs, and better access to computing power via cloud services. In addition to that, there is cheaper storage and analysis of data through the availability of targeted databases, software and algorithms. On the demand side, financial institutions have incentives to use AI as an opportunity for risk management gains, cost reduction and productivity improvements, which lead to higher profitability (Dapp and Slomka, 2015). Financial institutions are searching for cost effective means of complying with regulatory requirements, data reporting, trade execution, and rules on anti-money laundering and combating the financing of terrorism (AML/CFT). Moreover, AI and machine learning are implemented for a number of uses across the financial sector, such as customer-focused (credit scoring, insurance, and client-facing chatbots), operations-focused (automation, capital optimisation, model risk management and market impact analysis), trading and portfolio management and regulatory compliance (Van Liebergen, 2017).
Scholars have found that technological innovation possesses the necessary resources to change the business model, in order to achieve competitive advantage (Laeven, Levine & Michalopoulos, 2015). However, the place of artificial intelligence in the business model has yet not been acknowledged. AI is a compelling information system advancement, which can benefit organizations in diverse ways (Russel & Norvig, 2016). Information systems adoption is a challenging process and the implementation of AI can be a lengthy and difficult operation on an organizational level (Marr, 1977). When information system adoption and corporate strategy are aligned, greater efficiency and effectiveness lead to a successful execution of corporate strategy (Chan et al., 1997). This research aims to assess the level of adoption of AI as an information system innovation and examine its effect on strategy implementation as a way of creating value for the organization.
2.2.2 Acceptance of artificial intelligence
Owing to the great number of applications artificial intelligence has found, the requirements for future employees might change (Rifkin, 1996). Employees who perform simple and repetitive work will hardly be needed, which is already visible today, as the numbers of factory workers are constantly decreasing (Rotman, 2013). Therefore, many employees are hostile towards intelligent information systems. The reservations of employees are associated with fear of massive job cutbacks. Machines cost money only once, while labour costs are major costs for organizations and a recurring expenditure. In addition, machines and algorithms perform work with precision that is hardly achievable by humans. As an effect, humans can be considered inferior to machines in a competitive situation (Rifkin, 1996). Chen and Popovich (2003) state that governments and companies have to create general acceptance for such technologies, in order to make it transparent that the machines are not there to replace people, but to replace repetitive and autonomous tasks. When introducing new systems, employers have to proceed sensitively and gradually by establishing clear rules for working with the machines and specify the corresponding functions. They have to also be clear that the machine has only an assistive function and is not there to replace employees, as the power to make decisions still lies with the human being. Kalleberg (2009) also suggests that employees have to be involved in the development and the process of change at an early stage in order to grow accustomed to the new technology themselves. In addition, an adaptation of the education system is necessary, schools and universities have to encourage students’ interest in mathematics, information technology and science and teachers with digital competence can teach students how to think critically when using new media and help them achieve a fundamental grasp of new digital and information devices (Kalleberg, 2009).
Brynjolfsson and McAfee (2012) believe that technology boosts productivity and makes societies wealthier, however they also acknowledge the dark side of technology: it is eliminating the need for many types of jobs and leaving the normal worker worse than before (Tarafdar, Gupta and Turel, 2013). This has resulted in reluctance on the employees’ side to use the technologies that are being adopted,
which permits the organization from exploiting the innovation to its full potential. Therefore, artificial intelligence is not adopted fully, and this permits its full realization on strategic implementation, which as an effect does not create value. Therefore, for the purpose of this research we argue that artificial intelligence can create value for an organization only when it is accepted internally.
Combining the above leads to the final proposition: an organizational strategy that has clear goals, is data-driven and is accepted by the employees which as a result enhances the adoption of artificial intelligence. The goal of this research is to determine the most important capabilities needed for a financial organization to adopt artificial intelligence and to find out whether the adoption of artificial intelligence is successful only when it is accepted internally.
2.2.4 Technology Acceptance Model
Despite the impressive advances in hardware and software capabilities, information technology adoption and use in the workplace remains a central concern for research and practice. Sichel (1999) states that a low usage of installed systems has been detected as one of the major factors causing the “productivity paradox”. Therefore, understanding the conditions under which information systems are embraced by the human workforce in the organization is a high-priority research issue (Venkatesh and Davis, 2000). The major findings in the academic literature derive that the Technology Acceptance Model (TAM) by Davis (1989), presents a reliable conceptual framework for technology acceptance evaluation. TAM hypothesizes that an individual’s behavioural intention to use a system is regulated by two assumptions: perceived usefulness, defined as the extent to which a person believes that using the system will enhance their job performance, and perceived ease of use, defines as the extent to which a person believes that using the system will be free of effort (Davis, Bagozzi and Warshaw, 1989). The effects of external variables, such as system characteristics and facilitating conditions, on intention to use are mediated by perceived usefulness and perceived ease of use (Figure 2). According to the model, perceived usefulness is affected by perceived ease of use. This is illustrated by the assumption that the more effortless the system is to exploit, the more useful it can be (Venkatesh and Davis, 2000).
Figure 2. The Technology Acceptance Model (TAM). Davis et. al. (1989), Venkatesh et. al. (2003)
The Technology Acceptance Model proposes that external variables, such as the characteristics of the behavioural target, influence behavioural intentions only indirectly by “influencing the individual’s beliefs, evaluations, normative beliefs, motivation to comply, or the importance weights on the attitudinal and subjective norm components” (Fischbein and Ajzen, 1975, p. 307). External variables encompass all variables not explicitly represented in the model, and include demographic or personality characteristics of the actor, the nature of the particular behaviour, and persuasive communication.
Up to this date, there is no available literature on the adoption of artificial intelligence and its relation to strategy implementation. The purpose of this research is to evaluate the acceptance of artificial intelligence on an organizational level as a prerequisite determinant of the effect of artificial intelligence on the implementation of strategy. The Technology Acceptance Model has accumulated substantial empirical and theoretical support and is one of the most cited models in Information Systems, because due to its universal character it can be applied to any kind of software (Davis, 1989). The TAM provides reasons for the adoption of a technology and explains variety in the corresponding technology (Wallace and Sheetz, 2014). Therefore, for the purpose of this research the TAM model will be used to evaluate the acceptance level of artificial intelligence.
2.3 Conceptual model
From the literature review the following proposition is derived (Figure 3). It extends Davis’ work into the area of artificial intelligence, examining the relationships between the characteristics of the information system and organization characteristics as determinants of the adoption of artificial intelligence.
Figure 3: Modified Technology Acceptance Model: Artificial Intelligence Acceptance
2.3.1 System Characteristics
Characteristics of the innovation have been identified as perceived by the adopting organizations as having and influential effect on innovation adoption (Rogers, 2010). According to Rogers’s (1983) innovation theory, “an individual form an attitude toward the innovation, leading to a decision to adopt or reject and, if the decision is to adopt, to implementation of the innovation” (p. 48). Thong (1999) argues that the perception of the adopter towards the information system to be implemented is the primary determinant of the adoption of this system. Tornatzky and Klein (1982) have identified three innovation characteristics, prominent to the attitude formation process: relative advantage, compatibility and complexity. Relative advantage is described as the degree to which an innovation is perceived as superior to its antecedent, and the more positive perceptions of the benefits of IS, the more incentive for the organization to adopt the innovation (Rogers, 1983). Compatibility is referred
to as the degree to which an innovation is perceived as compatible with the existing needs, values and past experiences of the adopter (Rogers, 1983). Rogers (1983) further argues that organization are more likely to adopt innovations that are coherent with their existing work practices. Lastly, complexity is defined as “the degree to which an innovation is perceived as difficult to use” (Rogers, 1983). Therefore, the complexity of the information system is expected to influence the adoption decision (Thong, 1982).
2.3.2 Organizational characteristics
According to Thong (1982) the level of acceptance of an innovation is also determined by the characteristics of the organization in which it is adopted. The author determines three organizational characteristics, which influence the attitude formation process: business size, employee’s IS knowledge and information intensity (Thong, 1982). Scholars have found that larger businesses have more resources and advanced infrastructure to facilitate innovation adoption, while small businesses often experience a lack of resources, due to highly competitive environment, financial constraints and lack of professional expertise (Ein-Dor and Segev, 1978). Due to these constraints, small businesses face lack of IT expertise and management perspective, which builds barriers to technology adoption (Welsh and White, 1981). As a consequence, small firms are less likely to adopt IT as compared to large businesses (Ettlie and Bridges, 1982). Furthermore, because of the obstacles with developing the necessary skills and knowledge to apply the new innovations, businesses are pressured to postpone adoption until they have the necessary internal expertise (Dess and Robinson, 1984). In addition, there is empirical evidence that businesses with employees who have more knowledge of the technological innovation are more likely to use the innovation (Ettlie and Bridges, 1982). Hence, if employees are more knowledgeable about IS, the businesses may be more willing to adopt the IS innovation (Thong, 1982). Similarly, research has found that information intensity also has an effect on IT adoption (Wold, 1982). According to Yap (1990), businesses in different sectors have different information-processing
requirements, and those in more information-intensive industries are more likely to adopt IT than those in less intensive sectors. Online payment providers, for instance, are more information-intensive, as their main functions are to process and package payment information and cardholder data. Therefore, the higher the information intensity, the higher potential for strategic uses of IT in an organizational environment (Porter and Millar, 1985).
The modified TAM model has the objective to improve the understanding of user acceptance processes, providing new insights into the successful design and implementation of information systems, in particular artificial intelligence. In addition, the model should provide a theoretical basis for a practical “user acceptance testing” methodology, which would permit information system designers and strategy implementation practitioners to evaluate new proposed systems before their adoption (Davis, 1980).
2.3.3 Usage behaviour
The Technology Acceptance Model theorizes that an individual’s behavioural intention to use a system is determined by: perceived usefulness, referred to as the extent to which a person believes that using the system will enhance their performance, and perceived ease of use, the extent to which a person believes that using the system will not be complex. According to the model, perceived usefulness is also influenced by perceived ease of use, because “other things being equal, the easier the system is to use the more useful it can be” (Venkatesh and Davis, 2000, p. 187). In empirical tests of TAM, perceived usefulness is established as a strong determinant of usage intentions, which is why it is important to understand the determinants of this construct (Venkatesh and Davis, 2000). On the other hand, perceived ease of use has exhibited a less consistent effect across empirical studies, which is why the determinants of this concept have not been researched extensively (Venkatesh and Davis, 2000). Venkatesh and Davis (1996) state that a better understanding of the determinants of perceived usefulness and perceived ease of use would provide the opportunity to design interventions within the
organization, with the main goal to increase user acceptance and actual usage of the implemented system. Therefore, this research aims to extend the TAM to include additional key determinants of perceived usefulness and perceived ease of use in order to understand how the effects of these determinants affect overall usage behaviour.
All of the above comes together in the conceptual framework. Organizations create value in order to survive. Artificial intelligence allows firms to make strategic decisions based on data, which can create value for organizations (Russell & Norvig, 2016). The adoption of artificial intelligence is required for the technology to be used at its full potential and thereby create value. Thereafter, organizations adopt artificial intelligence to create value. The main challenge in the adoption process of artificial intelligence is the organizational readiness to adopt. The six different external characteristics (relative advantage, compatibility, complexity, business size, employee knowledge and information intensity) are examined to see how the adoption of artificial intelligence can be influenced.
III. Research design 3.1 Research strategy
A research strategy is established to achieve the research objectives and ensure the quality of the research. The research design of this paper is qualitative and is performed by means of semi-structured interviews with open ended questions. The Technology Acceptance Model is evaluated by qualitative means, which is also in line with the exploratory character of this research, as no research has previously assessed the model through a qualitative research method. Former research on the TAM has been conducted by means of quantitative methods, which focused on identifying factors that have a significant influence on technology acceptance but discovering ways to increase it was not the goal (Vogelsang, Steinhüser and Hoppe, 2013). A main limitation of analysing the TAM in a quantitative way is that it does not provide the opportunity to explore the background and the details of the research topic, as it is strictly limited to the relationships between the variables (Vogelsang et al., 2013).
Vogelsang, Steinhüser and Hoppe (2013) state that “this makes the authors unable to comprehensively analyze complex relationships like the interaction between man and technology” (p. 6). The authors propose a qualitative approach to the TAM, where with the help of qualitative data the researcher will be able to describe and measure the relevant factors of software acceptance and manipulate them (Vogelsang et al., 2013). The authors further state that a qualitative approach to the TAM derives coherent and comprehensible constructs and suggests steps on how to progress, because the direct dialogue between the researcher and the specialists in the field allow for further inquiry into experiences and best practices (Vogelsang et al., 2013). This is relevant for this research, as establishing the factors that affect the acceptance of artificial intelligence requires a deep examination of the attitude of people towards the technology. In order to be able to reach the research objectives, the data of this exploratory research is collected through a qualitative research method, following the research ideas of Vogelsang et al., (2013).
3.2 Data collection technique 3.2.1 Semi-structured interviews
Interviews are the most popular primary data collection method with exploratory studies as they can help gather valid and reliable information that is relevant to the research question and research objectives (Saunders, 2011). According to Saunders and Lewis (2012), in-depth interviews are the most commonly used method in qualitative research, because they are based on the personal experiences and beliefs of the participants. Yin (2009) states that the main advantage of in-depth interviews is the opportunity to collect more detailed information on the researched subject as compared to other data collection methods because they reflect interviewee’s thoughts and intentions about the research topic. Moreover, due to the exploratory nature of this research semi-structured interviews will be conducted, in order to identify possible additional factors that can influence the adoption of artificial intelligence. The choice of semi-structured interviews over structured interviews
has been made because of the opportunity to ask additional questions during the interview collection process, as new insights might be discovered (Saunders and Lewis, 2012). Semi-structured interviews are used to gather rich, descriptive data. This method provides a balance between the flexibility of an open-ended interview and the focus of a structured interview and uncovers valuable information on the personal experiences of participants (Doody and Noonan, 2013).
3.2.2 Purposeful sampling
For this research purposeful sampling is used, where a total of eight managers were interviewed. Purposeful sampling is used to select information-rich participants from which one can learn the essential perspectives, views and issues, which are important for the purpose of the research (Coyne, 1997). The eight respondents were purposefully selected because they are in managerial positions of successful start-ups and large corporates in the financial sector, currently working with artificial intelligence. The participants were considered suitable for this research because of their extensive knowledge and experience in the area. The variety of people being interviewed contributes with multiple viewpoints to work with in order to achieve the research objectives of this research.
IV. Data Collection
In this section the data collection process is explained. To begin with, a description of the sample is given in section “4.1 Sample”. In the subsections the selection criteria of the participants are stated (4.1.1), followed by a description of the sample of this research (4.1.2). Afterwards, the data analysis is described in section “4.2 Data Analysis”.
4.1 Sample
For the purpose of this research data is collected through eight semi-structured interviews on the uses of artificial intelligence in the financial sector. In order to select the interviewees, certain selection
criteria were applied. These criteria are elaborated in the following section. Subsequently, a description of the selected respondents is given, followed by an elaboration of the data collection techniques.
4.1.1 Selection criteria
The respondents are selected based on three selection criteria. Firstly, the respondents have to be working in an organization within the financial sector that is involved with the implementation of artificial intelligence into current processes. This is important, because artificial intelligence has to be relevant for the organization in order to be suitable for the purpose of this research. Secondly, the interviewees have to be in a managerial position within the company and are in charge of a certain group of employees or tasks. Managers of all levels were invited to participate, both lower-, medium- and higher-level management. Lower-level managers are able to see how artificial intelligence directly affects the performance of their team or department, which provides insight on the acceptance of AI and the uses of it within the organization. On the other hand, medium- and higher-level managers are able to provide information regarding the strategic processes and the role artificial intelligence plays in them, which is the centre research objective of this Master thesis. Moreover, managers are also the ones who have decisive authority and have the ability to make changes that are relevant for the organization as a whole, which presents valuable observations on the penetration of artificial intelligence into the strategic processes. Thirdly, in order for the findings to be transferable, there is a mix between technical participants (data scientist; technical support engineer) and respondents who have a more commercial role (consultant; sales executive). In addition, different types of organizations within the financial sector are to be represented, to reflect the variety of applications of artificial intelligence across the industry.
4.1.2 Sample description
In order to obtain a deeper understanding of the adoption of artificial intelligence into financial institutions and the relevance of AI on strategy, eight in-depth interviews are held with practitioners
within the field, representing the financial sector. Two of the interviews are conducted at the largest FinTech company in the world (Adyen), one at a multinational banking corporation (ING), one at a global payment method organization (TrustPay), one at a venture capital firm (Fenox Venture Capital), and one at a multinational consultancy corporate (Ernst & Young). One of the respondents is the Director of ISMC, a company developing modern applications for organizations wanting to adapt artificial intelligence and another one is the Founder and CEO of Forever Identity Inc., a platform that is capable of preserving one’s identity, biography, memory and physical aspects. Both of the respondents have a big portion of their clients coming from the finance sector, which makes their expertise a valuable input for this research. One of the interviewees (ING) is not in a senior position, however they work with AI on a daily basis and have a relevant overview of the effect AI has on their department. All of the interviewees have a postgraduate degree and they are coming from different educational backgrounds, both technical (5 respondents) and non-technical (3 respondents). This provides different points of view in both the technical implementation of AI and also the commercial aspect of implementing the technology. Furthermore, the management experience was a difference between the participants. The least experienced manager had a management experience of less than a year at the time of the interview, while the most experienced interviewee had over 15 years of management experience. An overview of the interviews is presented in the table below.
Figure 4: Overview of the respondents
4.1.3 Data collection methods
Access to each practitioner selected was obtained through a combination of phone calls and emails, using the autoher’s private network when appropriate. Some practitioners were approached via the professional network LinkedIn and were chosen based on the organization and the position they were currently occupying. Upon contacting the respondents, the topic of this research was introduced as “the relevance of artificial intelligence on the financial sector”. All participants were enthusiastic about participating in this research, and all of them regarded the topic as important to discuss. Not all of the practitioners contacted for an interview wanted to participate in the research, a total of four people refused to cooperate due to inability to set up an interview. For conducting the interviews, an appointment was scheduled at their offices. Five of the interviews are conducted face to face and three
through Skype, due to the fact that the participants were located in different countries at the time of the interviews.
An interview protocol was developed in order to ensure that the relevant propositions in the available literature were discussed. The protocol served as a guide in order to collect similar types of data from all participants while preventing logical gaps (David and Sutton, 2004). The structure of the interviews was based on the literature review, which provided analytical categories for the subsequent analysis (Pope, Ziebland and Mays, 2000). The interview protocol can be found in Appendix 1.
The interviews were flexible, with open-ended questions and the chance to explore topics that arose unpredictably depending on the direction of the interview (Doody & Noonan, 2013). The interview questions reflect different aspects of artificial intelligence adoption and the diverse uses of the technology across the organizations. Each respondent was interviewed in a period between 40-60 minutes. The interviews were conducted in Amsterdam, London and Singapore throughout the months of April and May. All of the interviews are conducted in English. No data is used without the participants’ consent and anonymity was provided.
4.2 Data Analysis
In order to make the research transparent, the interviews were recorded and afterwards transcribed. An example of one of the transcribed interviews can be found in Appendix 2. The remaining seven transcriptions are available by the author. Transcription is an important step of preparing the data prior to analysing and categorizing it, so that a consistent comparison can be disposed (Pope et al., 2000). The collected data is analysed through a computer-assisted qualitative data analysis software called Nvivo 11 (Yin, 2009). Through this software data collected through interviews can be thematically structured. This provides an opportunity to identify, analyse and see patterns within the data (Braun and Clarke, 2006). The analysis occurs by organising the data in codes. According to Miles and Huberman (1984), coding is necessary in order to reduce the large amounts of information into more
structured output. A code is a “a word or a short phrase that symbolically assigns a summative, salient, essence-capturing, and/or evocative attribute for a portion of language-based data” (Saldana, 2012, p.3). Despite the fact that this research is of a qualitative nature, this is a quantitative aspect, as the numbers of times a code appears within the data, reflects the importance of the code. As an initial step of the analysis, a word cloud using Nvivo 11 was derived. The word cloud provides a broad overview of the most used words and the associated topics and categories related to these words. The world cloud is presented below:
Figure 5: Word cloud: Most used words by the participants in this research
4.2.1 Coding
In order to code the collected data a combination of deductive and inductive approach is used. Through the deductive approach, a start list of codes is created based on the existing literature. In addition, new emerging codes and themes are added throughout the analysis of the collected data, which resembles the inductive approach. This combination enables to reformulate the predefined categories in order to find meaningful patterns in the data (Saunders and Lewis, 2014).
In the first step thе unit of аnаlysis is dеfined, which refers to the basic unit of text that is clаssified during the data analysis. According to Weber (1990), mеssagеs have to be utilisеd upon coding, and minor diffеrеncеs in the unit definition might affect coding decisions as well as the comparability of outcomеs with other studies. Therefore, defining the coding unit is considered as one of the most fundamental steps of the qualitative analysis (Weber, 1990). Quаlitаtive content аnаlysis uses individuаl themes аs the unit of analysis, such аs a single word, a phrаse, a sentence or a pаragrаph. By using themes as the coding unit, the analyst is primarily looking for the expressions of an idea (Minichiello, Aroni, Timewell and Alexander, 1990). An overview of the codes is made visible in Appendix 3.
4.2.2 Analysis
After coding, the data is organized into main categories, which means that concepts that belong to the same category are grouped together. According to Saldana (2013) this allows to move from multiple codes to a few concepts. Subsequently, a cross-case analysis is performed in order to create an overview of the results. Cross-case analyses are used to search for similarities and differences across data sets (Saldana, 2013). By creating word tables, the data from the individual cases is displayed according to a uniform framework, the base for the data collection (Yin, 2009). The cross-case analysis tests the extent to which different groups of cases share similarity or not, it can go beyond the initially discovered patterns and reveal important information, because it provides the chance to see systematic patterns and relationships among the cases (Miles & Huberman, 1994). The cross-case analysis is performed through framework metrics in Nvivo 11 and consists of the individual cases and the varieties of codes that were derived from the data. The cross-case table contains very detailed information, which is why an example of the general overview of the results of the different cases is constructed (Appendix 4).
4.2.3 Credibility, transferability and dependability
The participants are requested to confirm the interview transcripts in order to ensure credibility. In order to ensure transferability, the interviews are conducted with respondents representing different companies, based in different locations (the United Kingdom, the Netherlands, Singapore and Slovakia). In addition, as mentioned previously a variety of roles is also represented, both technical and non-technical. Moreover, some of the respondents are working with different companies across industries, to help them in the adoption process of artificial intelligence, which further increases transferability. The dependability of this research could be jeopardised because the observer is currently working at one of the organizations, Adyen, represented by two of the respondents. However, as the rest of the interviews are with participants representing organizations that are not related to Adyen and the participants are located in different countries, this will not be an issue. The results of this data analysis will be presented and discussed in the following section.
V. Research results and discussion
The results of the data analysis will be discussed in this chapter. To begin with, the various uses of artificial intelligence (5.1) among the organizations of the participants are presented, including risk management (5.1.1), automation (5.1.2), decision making (5.1.3) and customer support (5.1.4). Thereafter, in chapter 5.2, the acceptance level of artificial intelligence is discussed, as determined by the characteristics of the system (5.2.1), the organizational characteristics (5.2.2) and the usage behaviour of the people working with artificial intelligence (5.2.3). The results section will end with chapter 5.3, which presents the relation of artificial intelligence to strategy implementation.
5.1 Uses of artificial intelligence
The respondents were asked to talk about the different applications of artificial intelligence within their organizations, and the extent to which they consider it beneficial for performing their job. This
allowed for a comparative evaluation between the different uses of AI in the different organizations. In recent years the applications of artificial intelligence have grown. Robotics are not only applied for replacing manual tasks in warehouses and cargo bay operations, the applications range from collecting, analysing, and making decisions to helping organizations improve performance trade-offs and achieve exceptional levels of quality and efficiency (Binner, Kendall and Chen, 2004). This is in line with the opinion of participant A:
“AI is an exciting field, that will become more and more applicable. It has so many uses and can be implemented into so many things.”
The variety of financial institutions represented by the respondents provided a good overview of the applications of artificial intelligence within the financial sector. From the results it became evident that artificial intelligence is highly compatible with financial businesses and institutions. In addition, the main aspect that emerged from the interviews is that AI is used for risk management, including fraud detection and monitoring. Other uses include using AI to predict trends, analyse data and make decisions. AI is also considered beneficial in task automation and decision making. Moreover, the use of artificial intelligence in customer support chats for institutions that deal directly with customers also emerged. An overview of these applications can be observed in the world cloud presented below.
5.1.1 Risk management
The majority of the respondents stated that artificial intelligence is used for risk/portfolio management. Risk management is defined as the identification, assessment and prioritization of risks in order to depreciate, monitor and control the probability of unfortunate events or to maximize the achievement of potential opportunities (Saunders and Thomas, 1997). Artificial intelligence is compatible in this area, because it involves computer platforms that have the ability to address complex situations that are designated by uncertainty and ambiguity (Wu, Chen and Olson, 2014). One such aspect of risk management is fraud detection. Fraud is the use of stolen credit cards, misleading accounting practices or forging checks in order to gain access to extrinsic funds (Kou, Lu and Sirwongwattana, 2004). Six out of the eight participants acknowledged fraud detection as one of the main applications of AI in the financial sector. Regarding the use of AI to combat fraud participant G stated the following:
“I would say the most common use of AI within the financial sector is fighting against fraud, whether it’s to combat fraudulent transactions, chargebacks on card payments or prevent submitting falsified documents by clients. Fraud in general is one of the biggest issues of all financial institutions that they have to deal with.”
Kou, Lu and Sirwongwattana (2004) state that fraud detection involves “monitoring the behaviour of populations of users in order to estimate, detect, or avoid undesirable behavior” (p. 749). The authors further elaborate that undesirable behaviour is a term used to describe fraud, intrusion, and account defaulting, which can be committed by means of using stolen credit cards and unauthorised attempts to access and manipulate information (Kou et al., 2004).
The challenge is that without the right tools and systems in place, even small percentages of
fraudulent activity can quickly turn into big losses (Kou et al., 2004). With the advancement of technology, criminals have become clever, which is why in often cases the traditional fraud systems
