7.1 Introduction

The current chapter explores the impact of Artificial Intelligence (AI) on employment in two different sectors, namely accounting and banking. AI is relatively new to accounting (ICAEW, 2018). Currently, only a handful of big firms make considerable use of it. However, the growing AI investment by accounting firms cannot go unnoticed. Therefore, the first study looks at the potential impact of AI on the supply of labour while answering questions like: are the accountants vulnerable to losing their jobs as AI takes over data analysis and other accounting functions? How do employability prospects, skills and learning change with AI? What is expected of accountants to remain competitive in the labour market? Addressing these questions will enhance our understanding of how accountants could mitigate risks and accommodate the emergence of AI.

Unlike the accounting sector, the use of AI in banks has been increasing for years, varying from clients’ experience and support to trading and data management. The second study queries the actual use of AI in banks’ recruitment and selection processes within this context. The main questions this second part of the fieldwork sought to answer were: do Cypriot banks use AI in their Human Resource processes? If yes, how is AI utilized in recruitment and selection? And, what are the alleged implications, either positive or negative? The analysis draws on data collected through online interviews. The sample consists of five senior Human Resource managers, each sitting at a different financial institution. It is worth noting that, together, these five entities had a combined market share exceeding 70% of retail deposits (Central Bank of Cyprus, 2021). These interviews were conducted online due to ethical and health and safety guidelines during the pandemic, which required researchers to avoid face to face interation and the need to maintain social distancing.

The two sectors, of course, are not examined as homogenous recipients of technological change. However, their disparity allows comparing and contrasting the narrative of AI (accounting) with its actual impact (banking). The remainder of this chapter is structured as follows: in the next section, the literature review is divided into two distinct parts. The first reviews literature concerning AI in the field of accounting. The second sub-section looks at the application of AI for recruitment and selection in the banking sector. The third section presents the data collection and analysis methods used during fieldwork. The final two sections present fieldwork findings and advance a tentative conclusion.

7.2 Technological Change and the Labor Market

7.2.1 Challenges of AI for the Future Accountant

The practice of accounting is multifaceted. Although these practices have co-existed with applications of computer and information technologies since at least the 1950s, there still remain several time-consuming and labour-intensive processes that need to be undertaken by qualified accountants. Hence, the different forms of AI may impact practice in a number of ways. For instance, the complexity characterizing the tasks of accountants and auditors, alongside the ever-changing environment, could benefit from expert systems (O’Leary, 1986).

One such expert system is so-called machine learning (ML), which is a subset of AI. ML is seen as useful to accountants due to its ability to collect and learn from data. A machine is able to use algorithms to break down data, learn from data and continues to learn (Shalev-Shwartz & Ben-David, 2014). Also, ML can be used to classify a transaction and forecasting (Shimamoto, 2018). For instance, in the United States, 48 million people completed their own tax returns online in 2014 using automatic tax-return software powered by ML applications (Susskind & Susskind, 2017, p. 85).

AI’s expert systems can be helpful for both small and large enterprises. For instance, the cash-flows, invoices, receipts and other tasks which are typically assigned by small businesses to accountants, are now undertaken online (Susskind & Susskind, 2017, p. 86). Also, for larger companies, tax systems are available which are capable of collecting data, calculating tax returns, annual accounts and other reports. In 2009, 70% of the FTSE-100 companies prepared and submitted their corporate tax returns through an expert system designed by accounting consultants Deloitte (ibid.). ‘Deloitte Revatic Smart’ allows companies to recover foreign VAT payments using optical character recognition software with little human interference.

Li and Zheng (2018) suggest that AI improves usage of time, efficiency, reduces errors, and improves the industry’s competitiveness. For example, ‘robot accountants’ can offer business solutions to clients, while accountants concentrate on decision making (Shimamoto, 2018). It is a hybrid process where AI works along with humans rather than separately (Shimamoto, 2018; Susskind & Susskind, 2017); a good example here is when tax returns are completed and submitted online, whereas tax-experts serve as advisors. Besides, some authors argue that AI is just not good enough to undertake an accountant’s full role (Marr, 2018; Taylor, 2018).

More recently, it seems that accounting companies have been experimenting with AI for a broader range of purposes. According to Zhou (2017), three of the ‘big four’ accounting firms are testing different approaches. Ernst and Young (EY) focuses on AI’s Return on Investment (ROI). PriceWaterhouseCoopers (PwC) have been using ‘sprints’ by offering clients a demonstration of a working system and then improving it. Deloitte has a team dedicated to innovation, which focuses almost 80% of its work solely on applications of AI to improve the accounting and auditing functions. Also, while attempting to facilitate and enhance an understanding of AI, the Institute of Chartered Accountants in England and Wales (ICAEW) offers guidance and training to stakeholders, including regulators, governments and policymakers.

But despite the innovations mentioned above, it remains the case that AI is still relatively new to the accounting profession. For some, the application of AI solutions in auditing and accounting remains in its infancy when compared to other industries and sectors (ICAEW, 2018). However, the increasing use of AI applications and the growing investment by big accounting firms beg the question, ‘what would be the impact of AI on the accounting profession?’. As discussed in the next section, some worries and uncertainties have been recorded through fears that the increase of AI may threaten job prospects as the skill set of auditors and accountants, in general, is disrupted (Li & Zheng, 2018).

7.2.2 The Possible Impact of AI on Employment

McGaughey (2018) questioned whether the advent of the Internet, robotics and AI would bring a ‘jobless future’. This question is not entirely new. There has been an ongoing concern about the effects of digitisation on employment; and whether people will become unemployed. Some evidence suggests a negative effect on employment and wages for low and middle educated industrial workers (Mann & Püttmann, 2017). Τhe employment to population ratio increases within the commuting distance of a new automation technology innovation. Similarly, Acemoglu and Restrepo (2018) contend that the effects on wages and employment with one robot to a thousand workers would decrease employment to population ratio and wages, but only slightly. However, if the ratio of robots to humans is set to quadruple by 2025, then the effects would be much more serious.

Evidence from a study in 17 countries between 1993 and 2007 by Graetz and Michaels (2017, 2018), claimed that the increase of robots increases total productivity and wages. Also, robots may be reducing low-educated employment and the number of hours worked. In the same vein, Kemeny and Osman (2018) find that in metropolitan areas of the United States, growth in tech employment offers benefits to non-tradable (areas of work where the product cannot be produced in one country and sold in another) workers real wages.

Moreover, based on a study of 702 occupations in the United States, out of which 42% were on the verge of digitisation, Frey and Osborne (2017) confirmed that the introduction of machine learning and big data change certain types of tasks. Cognitive and non-routine tasks are becoming more obtainable, which in turn changes the structure of the workforce. This finding is consonant with Graetz and Michaels (2018), who claim that low-educated jobs are non-susceptible to computerization, whereas vacancies may relocate to more creative and social intelligence occupations. Following these findings, Study 1 will further explore the use and impact of AI on employment in the accounting sector.

7.2.3 AI for Recruitment and Selection in the Banking Sector

Digital systems become increasingly necessary across a wide range of industries around the world. The examples include educational platforms (Efthymiou & Zarifis, 2021; Doukanari et al., 2021; Efthymiou et al., 2021), platform-based supply-chains in the cruise industry (Sdoukopoulos et al., 2020; Efthymiou et al., 2022, the digital economy (Batiz-Lazo & Efthymiou, 2016a, b, c; Efthymiou & Michael, 2016), big data and analytics in tourism (Efthymiou, 2018; Efthymiou et al., 2020), blockchain and robots in hotels (Efthymiou et al., 2019), to mention a few. However, Artificial Intelligence (AI) is characterised by some distinguishing aspects. AI is a computerized system that is programmed by humans to work and respond in human-like manners. It is an artificial mimicry of tasks and capacities that would otherwise require human knowledge (Ryan, 2020). But most importantly, it seems that AI has the ability to redefine what can be digitised.

Currently, all AI is ‘narrow’, meaning it can only do what it is designed to do through specific algorithms. Yet, narrow AIs are mostly much better than humans at the task they were made for (van Duin & Bakhshi, 2017). Therefore, AI systems can be programmed to specialize in certain tasks, such as the function of recruitment and selection in banks. At the same time, conventional sourcing methods, such as handing-in job applications, are increasingly falling out of favour. Their place is taken over by e-recruitment and other internet-based processes.

AI appears as a solution to problems occurring in the ever-changing business landscape (Javed & Kabir Brishti, 2020). For example, the banking sector can utilize AI to automate pre-screening in the hiring process. The platform can support employers in locating candidates, examining their CVs, gathering reactions to pre-screening questions and leading psychometric tests. Candidates can be granted, and remain in continuous communication with the platform, on a 24/7 basis (Flinders, 2018). Using AI in the recruitment process saves time to recruiters, and enables them to step in later in the recruitment process. In other words, recruiting managers do not need to invest time overseeing huge pools of candidates. Rather, they can just focus on interviews and closing offers.

Another innovation in AI is the use of chatbots to communicate with candidates. Chatbots facilitate individualized commitment with candidates. Similar to virtual personal assistants like Siri, and Google Now, chatbots utilize regular language to understand messages and react to them. What is more, AI may use algorithmic computations, and comparative processes, to recognize if a candidate is dishonest. That can be achieved through repetitive questions with different sequence and meaning (Wilfred, 2018). Email, SMS, social media, and messaging apps are correspondence channels that can utilize chatbot applications to ease correspondence among candidates and organizations without time and area constraints (Ibrahim & Hassan, 2019).

Moreover, another AI function for recruitment and selection, which has become increasingly popular, is the so called ‘Video Interviews’. Video interviews can take place remotely, save time, and help overcome certain challenges, such as the need for social-distancing requirements during the coronavirus pandemic. AI has its own role to play in this process. It can be used to break down non-verbal communication patterns and facial expression during video interviews. Additionally, human bias during interviews is reduced, whereas, the interviewer is able to examine the recordings at a later stage, and take more precise decisions (Ibrahim & Hassan, 2019). AI works without the inadequacies innate to people, like exhaustion, feelings, predispositions, and restricted time. Also, the increasing intelligence of these machines makes them able to distinguish irregularities and recognize different particulars among large bases of data (Wilfred, 2018).

What is more, AI reduces the need for external organizations specializing in sourcing talent, along with their costs (Wilfred, 2018). In addition, AI can help organizations discover passive talents. This is because it utilizes bots to scan the web for potential candidates, including websites, virtual gatherings and tech chat-rooms, as well as traditional social media sites (Oswal et al., 2020). The bots aim at discovering a great match for the work, based on the set qualifications and standards of the position. They also explore if a certain individual is potentially ready for a job change, making it possible for organizations to poach talents which are uncommon to discover (Iqbal, 2018).

Furthermore, AI-supported technology makes the recruitment process faster. Unlike the conventional HR methods, where it is hard to screen numerous candidates simultaneously, organizations can set up an AI-enabled digital recruitment system. An AI recruiting assistant can converse with thousands of candidates at the same time. It answers applicants’ inquiries, positions them depending on pre-characterized factors, and alerts them when a position has been filled (Iqbal, 2018; Tambe et al., 2019).

Nonetheless, while AI can enhance the procedure of recruitment and selection, it also comes with certain challenges. While AI systems are incredible when dealing with repeated assignments, their creativity and imagination are limited. The affectability or innovation that comes from having the sense of seeing, hearing and feeling, cannot be imitated in the most intricate of machines. The natural sensibilities that people have during interviews can never be reproduced in machines (Wilfred, 2018). While human knowledge can be expressed through algorithms, human feelings and virtues can never be incorporated into them (Wilfred, 2018; Ibrahim & Hassan, 2019).

Departing from this review, the analysis sets out to explore phenomena through two different studies. The first study, which is conceptual, focuses on accounting firms, to explore the impact of AI on employment. The second study, which is empirical, examines the potential application of AI for recruitment and selection in the banking sector. The next section presents the study’s research design.

7.3 Research Design

The current research draws on mixed methods (empirical and conceptual), within the research philosophy of ‘pragmatism’ (Menand, 1997; Rescher, 2000; Rorty, 2000). Pragmatism ‘offers a method for selecting methodological mixes that can help researchers better answer many of their research questions’ (Johnson & Onwuegbuzie, 2004, p. 17). While ‘mixed methods’ research usually combines quantitative and qualitative research techniques into a single study, ‘pragmatism’ is open to additional typologies, dimensions and considerations. As explained below, the current chapter constructs a mixed model, by comprising conceptual (Study 1) as well as empirical (Study 2) studies.

Study 1: To examine the possible impact of AI on the accounting professions, the study has reviewed literature published by experts in the field of AI. The review looks at the intersection of AI, employment and accounting, published within the last 10 years. The search located 15 articles through the University’s online library, 59 articles on ProQuest, 53 articles on SAGE, and several practitioner reports through random searches on the Internet.

The literature was examined, coded, categorized and analyzed using the NVIVO software. Use of the ‘word frequency’ and ‘matrix’ have been made throughout the research. For the ‘word frequency’ creation, there were ‘stop-words’ in place to avoid irrelevant words. There was also a restriction of 50 words, and words with three letters and over. Through coding and breaking down each code, an analysis of the data collection was able to take place. NVIVO facilitated a systematic review through ‘case classification’; the creation of a visualisation of various trends; and the creation of big paragraphs of analysis.

Study 2: To explore the use of AI for recruitment and selection in the Cypriot Banking sector, along with its possible benefits and challenges, online interviews have been conducted with banks officials. The interviews took place in Spring, 2021. The interviews were meant to take place physically, through face-to-face meetings. However, the coronavirus outbreak (SARS-Cov-2) resulted in several postponements, and eventually, the collection of findings through online interviews. The method of in-depth interviews has been selected to secure a good level of responses.

Gaining access and consent proved to be a very difficult process, due to the data-security protocols applied in banks. Indicatively, only 50% of the predetermined sample accepted the invitation. However, after the first interview, the responder was kind enough to introduce us to colleagues in additional banks. At the end, five individuals accepted invitations to participate in the study. Although the sample size may seem to be small, the participants work in five main Cypriot commercial banks, with a combined market share of deposits of over 70% (Central Bank of Cyprus, 2021). They are also experts in their field, as they hold managerial positions in Human Resource departments. Therefore, the sample is representative of the Cypriot banking sector. The fieldwork was conducted in line with international Research Ethics standards. Informed consent, confidentiality and anonymity were applied to all interviews.

7.4 Findings

7.4.1 Study 1: Views About the Impact of AI on the Accounting Profession

The first visualisation from NVIVO shows that the papers based on accounting had both optimistic and pessimistic views. In nine papers, AI and its effects on the accounting profession are seen as positive. In one paper, AI’s impact is seen as negative. The four papers under the optimistic node believe that AI is likely to have a positive effect on accounting. They also rule out any possibility of AI taking over the profession and marginalizing human employees. When the nodes were further analysed it was found that a significant amount of views suggest that the accounting profession is not at risk by AI diffusion, now or in the short term. In five out of 13 papers, optimism concerns the short-term. This included statements like ‘Don’t press the panic button yet’ (Nagarajah, 2016, p. 35) and ‘widespread adoption in business and accounting is still in early stages’ (ICAEW, 2018, p. 8).

7.4.2 Accountants’ Attitudes

Another theme from the accounting papers was the need for change of accountants’ attitudes. Under the ‘accountants’ attitudes’ node, there were 10 papers with 30 references. All papers support the notion that accountants should have the ability to change when it comes to AI. The ‘accountants’ attitudes’ node was then sub-coded into the ‘need to embrace AI’ and ‘types of personality’. The results suggest a clear instruction for change in the accountants’ attitudes, which cannot be ignored. ‘Need’ was one of the most used words for this node. Moreover, the percentages produced by NVIVO about individual nodes, reconfirmed the lack of articles exploring AI and its impact on the accounting profession.

7.4.3 Skills, Robots, Data, Future and Investment Nodes

The node ‘skill’ included discussions based around the need for humans to be creative to stay relevant as AI cannot be creative yet. While there are accountants’ routine jobs being taken over by AI, other skills work hand in hand with AI. Another node concerns ‘robots’. It was found that robots are discussed along with their possible impact on all professions. Since robots are only getting cheaper and safer to use, reports state that jobs will be susceptible to this technology, as well as AI, sooner than we think. The node ‘data’ included 10 papers with 26 references making it a popular topic amongst the papers. It is clear from the node that data is beneficial to professions and there is an increasing amount of data which can be utilized. Unlike AI, ‘big data’ was discussed in a positive way in regards to its use. The node ‘future’ for both the accounting and non-accounting was split into two categories, optimists and pessimists. A common theme from the pessimist nodes was that AI will have effects on humans lives but not knowing exactly how or when. For the optimists, for as long as humans adjust their skill set, AI will not have massive effects. The node ‘investment’ showed the lack of investment in AI in the accounting industry. Whilst larger companies have begun research on AI, the slow return on the investment seems to have reduced the intentions to spend further in the accounting profession.

7.4.4 Study 2: The Digital Realm

Initial discussions with participants were centred around the ongoing digitization and algorithmic diffusion in the banking sector. For instance, one participant explained the importance of encouraging customers towards digital means and electronic banking. She mentioned: ‘[e]lectronic banking uses different algorithms. One example is the chat box, where customers ask for clarifications or additional help. These days, the majority of customers prefer to make their transactions electronically’. In another example, a participant explained how algorithms offer to customers a statistical analysis for the amount spent per week, month and year. The overwhelming impression was that bank officials support fully the digital turn, and expect customers to do the same.

Then, discussions focused on AI. Participants seemed to be aware of this technology, as well as the way it works for hiring purposes in the banking sector. As one participant mentioned, AI can be applied for recruitment purposes in several ways: ‘It can create accurate job descriptions. By analyzing current employee’s characteristics, abilities and skills, AI can define what the actual needs of the company are. Such job descriptions are likely to attract candidates, who fit the job description’. In another bank, a participant explained how AI can be used to find not only active, but also passive candidates. With AI, the attempt to find the perfect match for specific positions is increasing. All these findings resemble previous studies, as discussed earlier in the second section where extant literature was reviewed (e.g. Iqbal, 2018).

Another participant suggested that AI can be used for screening: ‘All applications are examined [whereas], applications that do not meet the required criteria are rejected. Instead of interviewing hundreds of candidates, AI simplifies the process’. These findings are consonant with previous studies, supporting that AI is indeed capable of saving labour time and related costs. The same participant explained the use of AI through the chatbot before the interview. Candidates ‘can answer a set of questions, created by AI, so that the number of interviews are reduced’. At the same time, she added, the chatbot remains available on a 24/7 basis, for uninterrupted communication with candidates.

Also, a participant explained that AI can be used for minimizing human involvement. ‘It evaluates candidates through automated questionnaires, prior to the interview, and produces objective results. For example, if the interview is online, through a camera, AI can recognize if a candidate is lying about something by understanding the body language. This is also relevant to previous studies, and adds extra objectivity to the recruitment and selection process.

However, while participants are well informed about the benefits of AI, none of the five banks is currently using it for recruitment and selection. As presented below, the reasons for not using AI vary.

7.4.5 Reasons for Not Utilizing AI in the Cypriot Banking Sector

One of the reasons for not using AI in Cypriot banks concerns the downsizing and restructuring of the sector. ‘Unlike 20 years ago, the bank now hires a considerably lower percentage. Fifteen years ago, we were hiring three- to four-hundred employees per year. Now, we only hire thirty to forty employees. The recruitment and screening department consists by specialists in the HR field, who now have to deal with fewer applications’, a participant mentioned. Another participant added, ‘[t]hey have the desired abilities, skills, knowledge and experience to deal with day to day operations’. In the same vein, a participant emphasized the team of experts that is currently in charge of recruitment and selection. ‘Our hiring needs have been reduced. There is a team consisting of well-trained and experienced employees. They evaluate candidates based on their skills, qualifications and behaviour. If an appropriate candidate cannot be located, they reach out to recruitment agencies’. Therefore, in all five banks, HR employees perform the hiring tasks without utilizing AI.

These findings are indicative of the intense downsizing and restructuring of the banking sector, which took place in Cyprus in the last decade. Cypriot banks with hundreds of branches around Cyprus and abroad downsized severely after the collapse of the economy in 2013 (e.g. Efthymiou & Michael, 2016). Of course, one of the questions arising from such findings is, what is better at times of downsizing and long-term cost efficiency, teams of human experts, or automation and Artificial Intelligence? As a participant admitted, ‘…more time and employees are needed to study, analyse the CVs, and continue with all remaining hiring stages, including rejection to short-listing, first interview, further shortlisting, additional interviews and more’.

Moreover, the findings reveal that the banks are interested in a very specific type of employee. A participant explained that since the hiring percentage has decreased dramatically, the bank seeks candidates who are already specialists on their fields. For example, they screen Cypriot bank employees who worked in London, where the banking system is similar. Also, the bank’s existing employees are asked to approach people they may know, with required skills and characteristics for specific positions and vacancies. According to the participants, these methods consume far less resource, and work in a targeted manner. Similarly, a participant mentioned: ‘we do not announce vacant positions publicly’, but later admitted that ‘AI could help as obtain a variety of CVs, for both active and passive candidates’.

Three of the participants referred to the lack of necessary infrastructure to support AI operations. They also referred to the high cost for implementing and running AI, along with the required knowhow. In addition, two of the participants revealed their discomfort with using advanced technology. They also revealed that their uncertainty concerning the use of an unknown technology, and the possibility of it being implemented inadequately, could affect their performance negatively.

But overall, most participants admitted that AI could help in some ways. As a participant mentioned, ‘…if AI was used for creating job description and processing applications, biases would be eliminated and selection criteria would be applied more accurately. Also, the use of AI during the interviews would prevent possible biases and add sustainability to the process’. In another bank, the HR coordinator mentioned that ‘[t]he use of AI would minimise possible discrimination, and evaluate each candidate with the same criteria. AI reviews are more objective than human reports; therefore, the final decisions are likely to be more objective’, another participant added.

In another bank, the participant added: ‘[h]umans’ choices are subjective, their decisions are based on their beliefs, values and perception. For example, to me, a high school certificate is a key factor for selecting the best possible candidate. For another recruiter, the high school certificate could be useless. But I believe that less time will be needed in recruitment as AI will complete a major part of the hard work. So, recruiters will focus more on details and other important tasks. Currently, no other competitors use it in Cyprus and Greece’.

Such findings reveal the contradictory nature of participants’ expressions. On the one hand, participants argue for AI’s potential benefits, especially in terms of objective evaluations, cost reduction and efficient hiring. On the other hand, the same participants negate previous arguments. Equally important are the findings concerning ethical consideration in the next section.

7.4.6 Challenges and Ethical Concerns for Using AI in Recruitment

The use of AI raises some ethical concerns. According to a participant, some candidates may belong to an older generation and be unfamiliar with the use of advanced technology. The obligatory use of AI during an online interview could cause stress and anxiety. Therefore, their performance as well as their ‘equal opportunity’ are likely to be affected negatively. Similarly, a participant, added, ‘we would like to see clear guidelines by the European Union and the Cyprus Government concerning bias prevention and towards ensuring equal opportunities and fair processes’.

Another participant expressed a concern on whether AI is intelligent and credible enough to produce objective evaluations. ‘If we cannot establish trust, the impact on our mission will be negative rather than positive’. Another participant added, ‘we often trust our guts and feelings of intuition, regarding a candidate’s potential fit into the team, or, we rely on references submitted to us by our network. By using AI in recruitment, the human factor will be eliminated and other implications could occur’. In the same vein, a participant questioned whether AI is really capable of understanding body language, tone of voice and different dialects. For example, ‘[i]n Cyprus we have a different dialect from Greece. Is AI capable of recognizing Cypriot language expressions?’

7.5 Discussion and Conclusions

While examining the use of AI in accounting and banking, along with its potential implications on employment, it seems that our findings have generated more questions than answers. In Study 1, we identified a number of interesting points. Although the use of AI in the accounting industry is currently limited, bigger firms have already started to invest in its utilization. AI has great potential and may become prominent as it becomes cheaper and smarter. Among the different types of AI, Machine Learning (ML) and Artificial Neural Networks (ANN) seem to be the most likely to spread in the sector. ML is currently being used in accounting firms, but not in smaller firms. ANN is deemed a complicated form of AI, which is yet to be used in the accounting industry. However, sooner or later, AI will become part of the profession. Therefore, accountants will have no choice but to embrace it. In order to remain employable, humans should first learn how to work along with AI. Then, they have to focus on skills that AI cannot replicate, such as creativity. This is potentially an area of future research in the field of HRM, especially in terms of new skills, new job specifications and arising needs.

Furthermore, in Study 2, the participants agreed on the benefits of using AI for hiring. They emphasized the weaknesses of current hiring methods, prior to suggesting that AI is likely to enhance the objectivity of the hiring process. Such findings raise some interesting questions. For instance, what are the implications on employment for sectors that are yet to adopt AI? What is the impact on the practice itself? Moreover, while the participants argue AI’s potential benefits, especially in term of objective evaluations, eliminating discrimination and biases, cost reduction and efficient hiring, the same participants expressed contradictory views. They actually voided previous statements by supporting the idea that AI lacks intuition.

Also, another paradox has to do with the feasibility of Cypriot banks in the long term. According to the officials participating in the study, AI is unnecessary due to the ongoing downsizing and reorganisation of the sector. However, would not AI be helpful for a sector trying to reduce costs? Would it not be beneficial to have digital expert systems (along with human experts) at times of downsizing and cost reengineering? Moreover, the Cypriot banks are interested in a very specific type of employee, with existing skills and experience. Fundamentally, such findings position the organizations against the principles of proper recruitment and selection. Certain benefits, such as developing a diversified and inclusive workforce, are reduced. Overall, such contradictory expressions have blurred our understanding even further. Equally important are the findings about ethical considerations. However, the concerns were expressed by officials who have no direct engagement with AI, in a sector that lacks the necessary infrastructure and knowhow to operate such expert systems. Therefore, it seems that the concerns were expressed on a rather hypothetical basis.

Overall, the comparison between the narrative of AI in accounting, and its actual impact in banking, provide evidence that the use of AI is not as widespread as its narrative would lead us to believe. However, stemming from both studies are some potentially problematic implications for current employees and future job applicants. It seems that bank employees, even at managerial levels, are uncertain and ambiguous about the actual use and benefit of AI. At the same time, accountants are muddled concerning AI’s impact on their jobs. In closing, it seems that for certain sectors and tasks, AI remains a foggy shadowland.