How do employees in the Polish financial sector react to automation in their workplace?

The COVID-19 pandemic significantly expedited the integration of digital technology in the banking sector (Shabir et al., 2023). Faced with the necessity of social distancing and economic efficiency, banks and other financial institutions rapidly evolved into tech-driven entities. They tailored their services to meet individual customer needs, leveraging data analytics to innovate.

The health crisis catalyzed a swift and comprehensive adoption of digital features, tools, and services across global banking operations. This included enhancing contactless payment systems, raising transaction limits, digitizing traditional processes such as account openings with electronic signatures, and facilitating remote customer service, including virtual branch appointments (Demir & Danisman, 2021).

Customer behaviors, alongside the attitudes of bank staff and partners in public administration, underwent a marked shift during the pandemic (Rybacka, 2023). Bank leadership and security teams adapted their strategies, focusing on problem-solving and adaptability. This period saw an accelerated digital transition, a trend that had been gradually progressing for years (Druszcz, 2017). Obstacles to this rapid digitization were primarily operational, encompassing conflicting business objectives and some employee resistance. Nonetheless, the urgency imposed by the pandemic made digitization a strategic imperative, primarily to simplify operations and maintain staff safety. Notable advancements included the widespread implementation of electronic signatures for distant dealings, the digitization of document management, and the increased application of automation in functions such as public aid schemes, exemplified by Poland’s crisis shield initiatives.

At the onset of pandemics, the financial sector has been already highly involved with artificial intelligence (AI) technologies allowing for automation. The AI Industry Exposure measure, created by Felten et al. in 2021, ranks industries based on their potential engagement with AI technology. This measure shows that 6 out of the 10 most AI-engaged industries, according to the 4-Digit NAICS classification, are in finance. The basis of this measure is the connection between AI advancements in various categories and different skill sets; the authors used two key data sources: the Electronic Frontier Foundation’s AI Progress Measurement, which tracks AI advancements through research literature, and the O*NET database, which provides details on job requirements and tasks (Felten et al., 2021). Although this measure is tailored to the U.S. labor market, it can still offer some insights into the AI engagement of industries in other countries, despite not being directly aligned with their specific occupational compositions. As noticed by Hamdan et al. (2022), “While artificial intelligence has been implemented more broadly in certain sectors than others, the banking industry is one of the few that has shown a reasonable degree of acceptance and implementation of this technology”. Similar to other organizations, banks anticipate that AI will enhance their productivity, predictability of operations, and responsiveness to changing factors in their environment.

It is also important to emphasize that the pace of adopting new technologies in banks over the past decade has been accelerated by a changing competitive landscape: the emergence of fintechs and techfins, offering convenient and cost-effective solutions for consumers, based on innovative use of intelligent algorithmization and the benefits of platformization (Stulz, 2019). These new financial organizations tend to adopt automation technologies even more actively than traditional banks, experimenting with solutions that allow for the cost-effective elimination of human labor (Nicoletti, 2021).

Automation using AI is increasingly being integrated into standardized banking tasks such as loan processing, document sorting, data analysis (particularly big data analysis), and certain aspects of customer service (i.e., the use of chatbots) (Engel et al., 2022). A systematic analysis of the literature conducted by Königstorfer and Thalmann (2020) revealed that AI applications in banking are extensive, ranging from customer service activities, through business operations, to managing back-end infrastructure. Employing the same methodology, Lazo and Ebardo (2023) identified five primary functional areas within which banks are keen to implement AI solutions: customer service, customer relationship management (CRM), fraud detection, credit and risk management, marketing, payment services, back-end operations, and data collection. By leveraging artificial intelligence, banks can minimize lending losses, enhance security in payment processing, streamline compliance-related tasks, and refine their approach to customer targeting (Fares et al., 2023; Truby et al., 2020)

Based on a case study analysis of the functioning of Norwegian banks (Andersen, 2020), it was determined that advancements in digitization and automation directly translate into cost efficiency improvement. Savings resulting from the application of artificial intelligence in front office operations primarily stem from the development of conversational banking, utilizing digital channels and increasingly well-functioning and naturally conversing chatbots for frictionless contact with individual and corporate clients. The use of AI enables the delivery of better decision-making insights for clients in the area of trading and investment decisions. Banks are able to confirm client identities more quickly and effectively. In the middle office, artificial intelligence technologies intensify fraud detection, especially in securing against money laundering and risk management, while in the back office, they minimize losses resulting from improperly conducted credit underwriting (i.e., the assessment of creditworthiness) (Digalaki, 2022). The adoption of AI-based solutions also increases regulatory compliance (Aziz & Andriansyah, 2023). Overall, Hamdan et al. (2022) emphasize that AI significantly contributes to improving the financial condition of banks by lowering their operational costs.

A frequently cited example of a specific technology that has made significant progress in banking since the early 2010s is the software used for Robotic Process Automation (RPA) (Smeets et al., 2021). RPA is designed to automate cognitive tasks by replicating human actions such as logging in, extracting data, and performing calculations, all based on a predefined set of rules. With the integration of machine learning, RPA evolves into Intelligent Process Automation (IPA) or cognitive RPA, becoming increasingly adept at handling new tasks such as natural language comprehension, sentiment analysis, and decision-making, as well as working with unstructured data. In banking and other financial services, Robotic Process Automation streamlines routine operations and enhances customer service efficiency by automating tasks such as data entry, transaction processing (e.g., fund transfers, loan disbursements), and responding to customer queries (e.g., balance inquiries). It also supports compliance, risk management, and data analysis by facilitating activities like compliance reporting, fraud detection, and loan processing, ensuring accuracy and speed. While performing all these tasks, the software replaces monotonous and routine low-value work, thus serving as a kind of “virtual workforce” (Villar & Khan, 2021). The ultimate aim of automation is to reduce human participation in work processes to minimum.

This latest observation leads us to a crucial question regarding the impact of automation on the labor market. The current phase of technological development is characterized by an advancing automation not only of physical tasks, as seen during the first and second industrial revolutions, but also of cognitive tasks, facilitated by the application of technologies commonly referred to as artificial intelligence. This has reignited discussions about the risk of widespread technological unemployment. In the 1990s, works such as Rifkin’s The End of Work (1996) posited that technology and automation might eliminate mass employment, a theme reiterated in contemporary publications like Susskind’s A World Without Work (2020). This idea gained prominence following the publication of a notable paper by Frey and Osborne (2017), in which the Oxford researchers concluded that nearly half of U.S. jobs could be automated within two decades, particularly routine tasks in professions such as accounting or banking. Their approach, however, was widely criticized as overly pessimistic and methodologically flawed due to its focus on entire occupations rather than the individual tasks within them. The majority of labor market researchers refer to the approach proposed by Harvard economists Autor, Levy, and Murnane (2003), which advocates for assessing the risks of automation based on specific tasks, leading to a perspective that views machines as complementing rather than replacing human labor.

Overall, academic literature, including Acemoglu and Restrepo (2020), shows mixed evidence on automation’s impact on employment, suggesting job creation often balances losses. In the context of work automation in banks, this mechanism is well illustrated by the case of ATM introduction: although the number of bank tellers employed in individual bank branches significantly decreased, the technology simultaneously contributed to the reduction of operating costs of these branches. The total number of branches increased, and with it, the number of people employed as tellers in the banking sector (Bessen, 2015).

Still, numerous publications emphasized the importance of skills that work in tandem with technology, warning against a growing skills gap. For example, Ford’s The Rise of the Robots (2016) cautioned that this might not equally benefit all workers, as highly skilled workers are more likely to work alongside machines than lower-skilled ones. Brynjolfsson and McAfee (2011) emphasized that the labor market may start experiencing a shortage of jobs for individuals with average skills, which are easily replaced by machine labor. Essentially, the current discussion on the impact of automation on the labor market strongly highlights the concept of Routine-Task Automation (RTA): tasks that are primarily at risk are those historically performed by humans that can be codified into a sequence of procedures, or algorithmized. In the literature pertaining to labor automation, this characteristic of occupational tasks is referred to as routineness. It is important to note that routineness implies not so much the repetitiveness of tasks, but their standardizability. And indeed, such tasks are relatively numerous in the banking sector. Therefore, banking professionals should increasingly experience task encroachment from intelligent automation technologies: applications of AI are taking over more areas of professional tasks. Additionally, it should be noted that RTA Theory suggests a polarization in the job market, where middle-skill jobs consisting largely of routine tasks decline, while both low-skill, non-routine manual jobs and high-skill, non-routine cognitive jobs see relative growth (Goos et al., 2014). This shift has significant implications for employment patterns, wage structures, and the skills demanded in the labor market.

In this article, we endeavor to answer whether employees in the banking sector in Poland are experiencing progressive automation in their work and what factors shape their attitudes towards automation. At a more general level, we sought to determine whether banking employees fear automation, given the consensus in the literature regarding the threat to routine tasks. Generally speaking, to our best knowledge the literature lacks empirical research that specifically addresses this question for this group of workers. Therefore, in conceptualizing our study, we referred to articles that broadly analyzed employees’ attitudes towards automation.

Dodel and Mesch (2020a) analyzed workers’ perceptions of automation’s impact in the workplace using the Pew Research Center American Trend Panel 2017. The dataset captures the attitudes towards new automation technologies such as workplace automation, driverless cars, robot caregivers, and computer algorithms and the subjective perception of their impact on jobs and wages. The authors found out that only 4.6% of workers experienced a negative effect of automation in the form of job loss or wage reduction. The view of technologies’ negative impacts on work was more prevalent among workers in less well-paid jobs and with lower levels of human capital, as well as among older cohorts. A frequent use of the Internet was associated with a more optimistic view of the tech-related future of one’s career. On the other hand, Nedelkoska and Quintini (2018) argue that the workers most vulnerable to automation are the younger workforce members, as their employment typically involves basic jobs that are more prone to being automated.

Mulas-Granados et al. (2019) investigated the factors explaining the perception of future automation impacts on an individual’s work. Their analysis revealed that a more pessimistic overview is more often found among older and poorer workers. The positive perspective was related to higher levels of job satisfaction and educational achievements, and workers revealing such an attitude more often recognized the need of reeducation and retraining. As the sample covered workers from 11 countries, the authors were also able to determine that the workforce from countries with a higher degree of automation perceived the impact of automation of their work more negatively, while the ones from emerging markets were more “likely to have a more favorable view of automation” (Mulas-Granados et al., 2019, p. 5).

Rodriguez-Bustelo, Batista-Fouget, and Selavos (2020) studied the concerns, perceived opportunities related to automation, and preventive measures among Spanish workers. Their analysis revealed that the level of education, work complexity, and diversity of skills used are negatively related to the level of fear of automation and that workers with a higher education level are more likely to undertake preventive steps. They also found out that work complexity is negatively related to fear of automation. Interestingly, the higher level of fear does not push workers to prepare for the future upheavals on labor market.

Several research studies have explored the relationship between automation-related fears, job insecurity linked to technology, and individuals’ labor market positions. Specifically, individuals with higher levels of education and better-paying occupations tend to exhibit lower levels of fear and insecurity, as highlighted by Dekker et al. in 2017 and further corroborated by Innocenti and Golin (2022). The latter indicated that older workers may have less fear of automation, potentially due to more secure job positions. In contrast, a study by Dengler and Gundert (2021) suggests a U-shaped link between age and job insecurity. Ivanov and his team’s research (2020) provides some backing for the idea that younger people, possibly because of their familiarity with technology, are less worried about being replaced by machines in their jobs. Additionally, this research points to a possible difference in attitudes towards job automation between urban and rural residents, with those in non-metropolitan areas showing more concern, possibly due to their different experiences with technology. In general, the anxiety surrounding automation stems from worries over dehumanization, peer pressure, the potential for jobs to be automated, and individuals’ perceptions of their own professionalism (Ivanov et al., 2020). It also should be noted that according to Arntz et al. (2022), age does not significantly affect perceptions of technology-induced unemployment risk in Germany.

In pursuit of our goal to investigate how the pace of automation processes changed during COVID-19, we assume, based on existing literature, that the pandemic has accelerated the digitization of enterprises and, with it, employees’ experience of automation, not excluding the financial sector (H1). Furthermore, our aim is to identify factors associated with the degree of automation experienced along with factors determining expectations of future automation and fear of automation among workers in the Polish financial sector. The literature review suggested that automation predominantly affects routine job tasks and requires a high level of digitization of a company. Hence, we hypothesize that automation experience (pre-pandemic and during the pandemic), along with future automation expectations and fear of automation, will be higher among workers in more routine occupations (H2) and in more digitized workplaces (H3). As educated employees tend to perform less routine tasks, we hypothesize an inverse relationship between obtained education and the level of automation experience, expectations, and fear (H4). The relationship with a worker’s age is less straightforward. Younger workers often perform simpler and more routine tasks (Nedelkoska & Quintini, 2018). However, it is the older workers who often reveal more negative views on technology and automation (Dodel & Mesch, 2020; Mulas-Granados et al., 2019). Hence, we hypothesize that the younger workers experience more automation, but the older workers expect higher automation in the future and fear automation more (H5). Furthermore, we anticipate that higher levels of automation experience translate into higher automation expectations and fear (H6).

Data presented in this article was collected via an online survey conducted between October and November 2020 among Polish employees in the financial sector. The survey was a part of larger research project titled “The Future of Work in the Financial Sector,” carried on on behalf of and in collaboration with the Sectoral Council for Competencies of the Financial Sector. The Council actively supported the recruitment of the respondents. The construction of the questionnaire was informed by 10 pilot interviews with bank CEOs carried out in 2017, which focused on the effects of digitization in the financial sector, and a survey carried out in 2019 concerning the specificity of the Polish banking sector. However, in this article we present solely the analysis of the data gathered in 2020, which amounted to 172 questionnaires. The interpretation of the results was helped by the insights collected via 37 interviews with banking managers.

The data was based on convenience sampling. The sample covers slightly more men than women, as the percentage of men equals 52 percent. More than a half of responders fall into the 36-47 age group (56 percent), and almost everyone (96 percent) graduated from a certain level of tertiary education (bachelor or higher). Only 5 percent worked less than five years in the financial sector, with one third working in the sector more than 20 years. A half of respondents work in foreign banks (51 percent), 27 percent in domestic (Polish) banks, and the rest represent non-bank institutions, such as insurance companies, pension funds, or brokerage offices (22 percent).

Regarding a firm’s character, we differentiate between digital, mixed, and traditional companies. A digital firm is defined as a company that focuses especially on developing its online activity and services and/or aims to reduce the number of branches in favor of remote service channels. A traditional company operates primarily through a network of physical locations but may develop some basic online presence. Finally, a mixed company places equal emphasis on traditional services delivered through branches, as well as on online activities and services. The character of a firm was self-declared by its employees. Only about one-fourth of responders work in a digital company, while the remaining 73 percent occupy positions in either a traditional or a mixed firm. The largest share of workers (41 percent) hold the position of an expert or a specialist, one-third are employed in a managerial position, and 14 percent advise customers. Table A contains detailed information on the sample characteristics.

In this study, we elicit the subjective views of financial sector workers on automation processes in their workplace through a deliberately designed questionnaire. To answer our research questions, we ask the respondents a series of inquiries about their experience with automation, both past and present, their views on future automation and their apprehensions about it. Initially, we scrutinize the pre-pandemic automation experience. To do so, we ask the respondents to estimate the share of their job tasks that have been automated in the three years before the pandemic. Subsequently, we take a closer look at workplace changes observed during COVID-19, including the pace of task automation. We differentiate between significant, moderate, and no acceleration of task automation in the period between March 2020 and the moment of data collection. Finally, we explore how these experiences and changes correlate with respondents’ perceptions of future automation and their fear of automation. When inquiring about future automation expectations, similarly to automation experience, we ask the respondents to assess the share of their job tasks that they predict to be automated within the next five years. Workers’ fear of automation is measured with a 5-point Likert scale regarding the following statement: Due to the advancing automation and digitization of the financial sector, I feel anxious about my job. See Table B for the exact phrasing of questions along with the answers and their categorization.

Since we designed the questionnaire when the studies on workers’ perceptions of automation were still in early development, our questions on automation expectations and fear do not directly align with the existing scales. Yet they are closely related to some of the survey items developed later, such as those by Arntz et al. (2022). The timeframe, however, is kept shorter as the digitization processes in the financial sector progress faster than in other sectors.

Our first goal was to assess whether respondents had witnessed changes in their job responsibilities due to the implementation of new technologies (such as IT systems, specialized software, algorithms, and robots) in the three years prior to the survey. Among reported changes, we defined a “major automation” as instances where respondents reported experiencing automation affecting over half of their tasks. Conversely, a “minor automation” was noted when automation impacted less than half of their job tasks (see Table B). We also sought for the correlates of the levels of experienced automation in accordance with the literature review and stated hypotheses.

In the years leading up to COVID-19, only 18 percent of respondents noted a major transformation of their job responsibilities due to new technologies. Over half, 57 percent, declared that less than half of their task was affected by automation, while one in four financial sector employees (25 percent) reported no discernible impact whatsoever (see Figure 1.)

Figure 1.

A major automation of job tasks was predominant among the oldest group of workers (38.2 percent) and, intuitively, among workers in digital firms (29.8 percent). However, the relationship between age and automation experience is U-shaped – workers in the middle age group most often reported no automation. Furthermore, a prime illustration of the advancing robotization and automation can be seen in the role of the customer advisor. This position is particularly vulnerable to shifts brought on by new communication channels and evolving customer service practices. Our survey reveals that a substantial 25 percent of customer advisors have seen over half of their tasks automated. This figure is notably more than double the rate for board members and high-level managers, where only 11.5 percent reported a similar experience. Nevertheless, the high-level managers along with individuals holding master’s degrees most frequently reported minor automation (see Table C).

To test the hypotheses further, we performed the Pearson Chi-squared test of independence to analyze the association between the variables of interest. The analysis revealed a significant relationship between the level of automation experience and responders’ age and education and the firm’s character. Cramer’s V measure indicates rather low strength of association between the significantly correlated variables (see Table D).

As the statistically significant variables from Table D are ordinal or dichotomous, we calculated the Gamma measure to obtain the direction of association. The results align closely with the previous descriptive statistics. The obtained Gamma value indicates a positive relationship between the level of automation experience and age – older workers seem to have experienced more automation. Also, being employed in a “digitized” company is related to broader automation experience. The relationship with education is negative – the more educated workers declared less changes due to automation.

The survey data reveals that, similar to other sectors, the financial sector witnessed a transformation in work organization due to COVID-related sanitation requirements, leading to a significant transition to remote work. Prior to the pandemic, in the Polish financial sector, 67 percent of our survey respondents exclusively worked from their offices, with 23 percent occasionally utilizing remote work options and 8 percent using hybrid arrangements. However, following March 2020, nearly 90 percent of employees experienced a shift in their work approach, with one in three transitioning entirely to remote work. Only 9 percent of finance professionals (approximately the same percentage as those who worked remotely before the pandemic) continued to work permanently from the office (see Figure 2).

Figure 2.

The pandemic has significantly heightened the perception of automation’s rapid expansion in the financial sector. Notably, 61 percent of surveyed individuals acknowledged a hastened pace in the automation of tasks within their responsibilities during the pandemic. Of these, 28 percent experienced a marked intensification, while an additional 33 percent sensed a moderate acceleration (see Figure 3). This surge in automation was starkly noticeable when juxtaposed with the preceding three years, including the pre-COVID-19 era, wherein the general awareness of automation and technological shifts was comparatively small.

Figure 3.

In the COVID-19 era, the most automated tasks were those associated with task reporting and correspondence, as highlighted by 84 percent and 82 percent of the participants, respectively. Furthermore, a significant portion of the workforce has also observed enhancements in processes such as data and information analysis (79 percent), data visualization (78 percent), and work monitoring (76 percent), along with the automation of mundane, repetitive tasks (see Figure 4).

Figure 4.

The analysis of association between variables (presented in Table E) revealed that the level of automation experience is the only significantly and positively (Gamma = 0.52) related variable in the perceived pandemic-related automation expansion. It appears that the workers who experienced higher automation in the past, noticed a higher level of automation acceleration during COVID-19.

The workplace transformations observed during the pandemic reflect broader trends that have been developing over several years in the financial sector. Although these processes of task flow automation and optimization have been ongoing, the stringent lockdowns initiated by the government in March 2020 served as a catalyst, expediting these changes. Consequently, the pandemic not only hastened inevitable changes but also tested the adaptability of institutions and their employees to thrive in a newly digitized labor market.

To measure the automation expectations, we asked individuals about the share of current occupational tasks which, in their opinion, were going to be automated within the next five years (see Table B). The results reveal varying expectations for job automation, with most foreseeing minor automation (65 percent) and a one-third anticipating major automation. Only 3 percent of responders do not believe that any of their tasks could be performed by technology, which highlights the widespread recognition of intense digital transformation in the financial sector (see Figure 5.)

Figure 5.

The financial sector’s workers seem to expect the shifts towards automation to represent an enduring change in how the sector operates, as opposed to, for instance, a temporary pandemic-related adjustment. Indeed, only one percent of the respondents do not expect the solutions introduced as a response to pandemic to endure (See Figure 6). This perspective is also reflected by the positive correlation between their experience of automation and its acceleration during COVID-19, and the level of automation expectation within the next five years. Furthermore, the automation expectations are positively and significantly associated with a worker’s age and the digital character of a firm (see Table F).

Figure 6.

Additionally, we decided to investigate the determinants of the fear of automation. Concerns about job security due to automation and digitization in the financial sector were measured by responses to the statement, “Due to the advancing automation and digitization of the financial sector, I feel anxious about my job” (see Table B). Interestingly, a majority of employees in the Polish financial sector, 60 percent, do not feel threatened by these changes and do not agree with the statement. In contrast, 36 percent acknowledge feeling a sense of job insecurity in light of the sector’s technological advancements (see Figure 7.).

Figure 7.

The analysis of associations indicated that the nature of one’s occupation is the sole significant factor linked to anxiety over job automation (see Table G). Predictably, roles characterized by routine cognitive tasks evoke greater fear of being replaced by automation. Customer advisors exhibited the highest level of concern, while experts and specialists showed considerably less apprehension. On the other end of the spectrum, board members displayed the least worry. Their roles, largely composed of decision-making within the dynamic non-routine realm of business, appear less susceptible to the risks of automation (see Figure 8).

Figure 8.