Abstract
Introduction
The use of artificial intelligence (AI) continues to grow in various areas of economies across the world. Generative AI will contribute $2.6 trillion to $4.4 trillion annually to the global economy in the years ahead (Yee et al., 2023). Currently, 35% of all businesses use AI to various degrees, and 42% of other businesses plan to use AI in the near future (Dennison, 2023). Besides the increasing adoption of AI, new challenges emerge to mitigate the negative impacts. AI creates various societal concerns, and organizational leaders play a vital role in mitigating those concerns (Raisch & Krakowski, 2021). However, organizational leaders face challenges in balancing the competing interests of stakeholders. Balancing the organization's business objectives and responsible digital practices is also challenging (Yokoi et al., 2023). For example, data localization requirements often contradict the efficiency of globally distributed value chains. Similarly, ethical scrutiny of AI algorithms tends to slow AI-based project developments in organizations (Yokoi et al., 2023). The emerging leadership era increasingly focuses on how algorithms can extend the leaders’ capacity to navigate AI to perform the core responsibilities in the organizations. Such an augmented leadership approach refers to AI as a co-leader to serve the partner with algorithm-based suggestions in various leadership circumstances (Quaquebeke & Gerpott, 2023).
The coding process.
Increasing AI applications across industries has driven scholars to investigate the role of AI in various contexts. Still, ensuring the benefits against the drawbacks of AI remains a significant challenge for organizational stakeholders (Johansson & Björkman, 2018). Recent AI research in management focus on a range of topics including competitive advantage (Kemp, 2024; Krakowski et al., 2022); leadership behaviors (Tsai et al., 2022); digital leadership (Banks et al., 2022); organizational learning (Balasubramanian et al., 2021); leadership traits (Doornenbal et al., 2022); employment relationships (Varma et al., 2022); bias mitigation (Choudhury et al., 2020); perceived users’ value (Gregory et al., 2021); organizational design choices (Murray et al., 2021); individual morality (Giroux et al., 2022; Tóth et al., 2022); ethical and cultural implications (Wright & Schultz, 2018); and paradoxical automation-augmentation organizational outcome (Raisch & Krakowski, 2021). Extant research shows that leaders’ role will be vital in an advanced AI application era (Agrawal et al., 2017; Davenport & Kirby, 2016; Kolbjørnsrud et al., 2017; Wilson & Daugherty, 2018). However, a key area for the current research priorities is to explore the specific capabilities required for the collaborative human-machine era. AI-driven capability refers to the leaders’ capacity to think and act in human-machine collaborative ways to lead in an AI-driven organizational environment. Based on the financial services industry evidence, we explore the dimensions of leaders’ AI-driven capability for responsible AI adoption in organizations.
Because AI-driven organizations operate in a relatively uncertain and volatile environment, we adopt a dynamic managerial capability (DMC) (Adner & Helfat, 2003; Harris & Helfat, 2016) theoretical approach to explore leaders’ AI-driven capabilities. The DMC approach relates to “the capabilities with which managers create, extend, and modify the ways in which firms make a living” (Martin & Bachrach, 2018, p. 27). The DMC theory explains the relationship between the quality of managerial decisions under conditions of strategic change and its effects on organizational performance. The emergence of AI increases uncertainty for organizations. In a complex organizational context, dynamic capabilities (Schoemaker et al., 2018) and leaders’ aligned responsible behaviors (Samimi et al., 2020) are critical to addressing uncertainties. DMC is difficult to configure and deploy and often works in specific contexts (Helfat & Martin, 2015). Extant research has investigated DMC in ambidexterity, competitive dynamics, innovation, diversification, and strategic renewal (Helfat & Martin, 2015). In AI environments, AI-driven capabilities are vital for leadership to address emerging challenges and direct organizations on the right strategic path (Davenport, 2016; Wilson & Daugherty, 2018). However, there is a gap on what AI-driven capability dimensions will create new insights to develop DMC in an increasingly AI-driven organizational context. Thus, we seek to address the following research question: What are the leaders’ key capabilities required in an increasingly AI-driven environment?
While previous research stressed the role of leaders in an increasingly AI-driven environment, there is a gap in the understanding of the specific capabilities required for leaders in an AI environment. First, we contribute to the emerging leadership discourse by introducing AI-driven leadership capability from the DMC lens. The DMC perspective is ideally suited for this study because it focuses on how non-static resources can be applied to managerial decision-making in dynamic environments such as AI-driven environments (Wamba et al., 2024). The focus is on leaders having the agility and adaptability of decision-making resources to make the best use of the changing environment to maximize organizational performance. Second, our findings offer unique insights into three dimensions that impact upper-echelon leaders’ cognitive frame to make strategic choices: technical, adaptive and transformational. Lastly, echoing Hoch et al. (2018), our findings suggest that transformational leaders could be better suited to fundamentally rethink the organizational strategies, structures, and resources to adopt in the rapidly evolving AI environment.
The rest of the paper is structured as follows. First, we discuss the ethical concerns in the AI landscape and the role of leadership. Then, we clarify the theoretical underpinnings of DMC, followed by a discussion on methods such as sampling and data collection, data analysis, and trustworthiness. Then, we present the interview findings with identified themes. Lastly, we present the theoretical and practical implications while highlighting the limitations and future research avenues.
Literature Review
The AI Landscape
AI can induce individual, organizational, and societal risks, significantly offsetting the benefits of digitalization (Alt, 2018). At the individual level, AI is negatively impacted by privacy concerns, content, and product recommendations (Grewal et al., 2021). For example, voice assistants like Alexa analyze customers’ voices and predict the moments using AI technology. Another example is facial recognition-based payments that exacerbate privacy risks because the human face depicts various personal information, including gender, appearance, age, etc. Personalized information often leads to perceived information narrowing and privacy concerns, resulting in technological resistance (Li et al., 2021).
From an organizational perspective, AI-based products may impact companies’ reputations and profitability when AI-enabled chatbots do not deliver to expectations and create a trust gap between customers and companies (Yen & Chiang, 2021). Organizations also face workforce challenges when integrating AI technologies (Cheng et al., 2022). For example, organizations use AI-driven software to automate many rule-based and repetitive tasks in manufacturing, customer services and data entry to improve efficiency. However, such automation leads to downsizing and layoffs of specific roles, resulting in job losses (Kelly, 2023). Job displacement increasingly creates income inequality and negatively impacts society because of wealth concentration among those who develop and own AI technologies (Lu, 2023). AI also creates ethical problems from a societal standpoint (Zajko, 2022). For example, organizations use AI to monitor customers’ online activity to understand their behaviours and movements without explicit consent. Such monitoring creates a ‘surveillance society’, negatively impacting the societal sense of freedom and trust (Alt, 2018). Financial organizations use AI-based models for loan approval. However, AI-based models may benefit certain demographic groups over others if the algorithms suffer from biases. The biased AI-based models result in discriminatory outcomes and can perpetuate societal inequality. At societal levels, issues like AI fairness and discrimination, AI regulation, and moral dilemmas pose increasing challenges to AI governance (Wirtz et al., 2020).
AI and Leadership
Cloud-based data processing has emerged with technology, geography, and policy issues for organizations. Combined with AI, analytics, the Internet of Things, and virtual and augmented reality impacts business strategies (Davenport, 2019). Such an emerging AI landscape raises more ethical concerns for organizations. AI's ethical risks also increase because of inherent algorithmic complexities, rapidly changing external environment and probability-based decisions (Babic et al., 2021). AI's ethical concerns impact customers, suppliers, employees, broader stakeholders, and brands and reputations. Hence, organizational leaders must take responsibility for AI-driven ethical concerns (Davenport & Katyal, 2018). The 2022 Optus security breach in Australia revealed the lack of leadership responsibility in the organizations. The Governance Institute of Australia found that 34% of the respondents think senior organizational leaders are responsible for such situations. Studies show that 50% of respondents believe that data management policy failure led to this security breach, which is the terrain of senior leadership (Tong, 2022).
In the AI landscape, ethical concerns are related to accountability, fairness, transparency, and safety. Leaders play a vital role in addressing the ethical concerns in these areas. Accountability refers to the human liability for the AI's actions and outcomes. Human judgment aids AI in assessing the various effects on stakeholders. Leaders play a vital role in identifying the responsibility (Kompella, 2022), answerability and auditability (Leslie, 2019) in each stage of AI's design, implementation, and monitoring. AI's fairness refers to the unbiased application for concerned stakeholders. Organizational leaders safeguard AI's fairness by ensuring an equitable analytical structure, fair processes of an AI-based model, and reasonable features (Leslie, 2019). Transparency deals with the ‘Why’ question of AI-based processes. Leaders justify the outcomes of AI-based processes by explaining how the AI-based model works. Such explanations clarify the rationales behind AI-based processes, producing non-discriminatory, ethically permissible, and publicly trustworthy AI-based solutions (Leslie, 2019). Safety is a critical dimension of AI ethics that establishes human control over the AI system. The AI system is safe when it meets the designer's expectations in changing circumstances. Leaders must evaluate the transformative effect of AI's technical sustainability on individuals and society to produce safe and reliable solutions (Leslie, 2019). Digital leaders are called upon to successfully manage the digital transformation and the organization in an increasingly AI-driven environment (Klein, 2020).
Digital leadership is an emerging construct that broadly encompasses leading both the digital transition and the organization in a digital environment. Digital leaders take the right initiatives and actions to manage digitalization in the organization (El Sawy et al., 2020; Porfírio et al., 2021). They deploy the right skills (Singh & Hess, 2020) to positively influence the organizational members affected by digital technologies (Avolio et al., 2014). These qualities involve specific competencies related to both technology management and strategic management. Previous digital leadership researches highlight leaders’ awareness of technological developments, utilizing technologies to harness opportunities and neutralize threats, managing the impact of technologies on organizations, products, and services, and dealing with employees effectively during technology-driven uncertainties in the organization (Kane et al., 2019; Kiron et al., 2016; Philip & Gavrilova Aguilar, 2022). Digital leaders act as role models by fostering changes and supporting employees to gain digital know-how (Leavy, 2020). However, considering the rapid disruptions of automated technologies, digital leaders must be visionary, tech-savvy, adaptable, highly collaborative, motivating, and creative (Day et al., 2014; Leavy, 2020; Reck & Fliaster, 2019).
The Dynamic Managerial Capability View
Strategic management literature has sought to synthesize insights from human psychology to understand the changing nature of the competitive industry structure (Peteraf & Shanley, 1997) and to clarify the nature of cognitive biases in strategic decisions (Bateman & Zeithaml, 1989). Organizations’ focus shifts from the external environment to internal capabilities as leaders develop organizational dynamic capabilities (Harreld et al., 2007; Hodgkinson & Healey, 2011; Ochie et al., 2022). Dynamic managerial capability is the dynamic capability that develops an organization's capacity to re-configure its existing resources and capabilities to sustain competitive advantage (Ambrosini & Altintas, 2019; Heubeck, 2023; O’Reilly & Tushman, 2008). Strategy scholars are paying attention to the individual-level cognitive and behavioural processes to promote learning, organizational adoption, and performance (Helfat et al., 2007). Dynamic managerial capabilities enable firms to differentiate strategic performances in the competitive landscape. Leaders use dynamic managerial capabilities by distinctively re-configuring organizational resources to translate into unique strategic and performance outcomes (Beck & Wiersema, 2013). From a micro-foundation perspective, dynamic managerial capabilities are the most influential theoretical lens in capability research (Guenduez & Mergel, 2022; Helfat & Peteraf, 2015). Micro-foundations are context-specific (Felin & Powell, 2016); hence, a review of capabilities is required for developing effective leadership in AI-driven organizations in the emerging AI environmental context (Felin et al., 2012).
Rosenbloom (2000) first mentioned the central role of leadership in the dynamic capability process to ensure organizational survival. Later, Adner and Helfat (2003) focused on managerial capabilities to create, integrate and reconfigure organizational resources in a dynamic capability process. Dynamic managerial capabilities emanate from three inherent factors: human capital, social capital and cognition (Helfat & Martin, 2015). These factors, individually and in combination, influence the organization's sensing, seizing and reconfiguring capacity. First, human capital is developed by managerial knowledge, skills and expertise. Human capital justifies differentiated performances in similar challenging circumstances. More diverse and complementary human capital positively impacts organizational performance (Wright et al., 2014). Second, managerial social capital explains how some managers efficiently exchange and combine organizational resources to impact performance positively (Ambrosini & Altintas, 2019). Organizations use managerial relationship networks to create access to unique knowledge to achieve business objectives (Hernández-Carrión et al., 2017). Third, managerial cognition is closely related to individual beliefs and knowledge, and it involves acquiring and processing information through mental activities (Colman, 2006). Cognition is complex to understand because human thoughts and ideas can not be observed directly. However, human thoughts reflect actions and behaviours, making the study of managerial cognition possible (Taylor, 2005).
Understanding dynamic managerial capabilities in AI environments allows managers to set priorities and determine the strategic factors for higher organizational performance (Helfat & Martin, 2015). In an increasingly AI-driven environment, dynamic managerial capabilities enable leaders to harness evolving technological advances. Leaders with such capabilities can use AI to enhance decision-making by producing actionable insights from predictive analytics, thus enhancing strategic decision-making. Leaders can use AI to dynamically re-configure and optimize human, financial, and technological resources to streamline operations by reducing inefficiencies. Leaders can also stay ahead in the competition by dynamically using AI to simulate various scenarios and assess potential outcomes. Moreover, the dynamic use of AI promotes agility by re-configuring organizational operations and strategies. Dynamic managerial capabilities also allow organizations to use their resources to neutralize threats and harness opportunities in competitive environments. Besides, the increasing trend of AI adoption creates ethical concerns (Davenport, 2019) and distinctive leadership challenges for organizations (Kompella, 2022). Dynamic managerial capabilities are vital in responding to those challenges, specifically emergencies, crises, and other disruptive circumstances (Ishida, 2020).
Research Methods
Interview Questions and Participants
We reviewed industry reports, magazines, newspaper articles, and scholarly databases to develop an original research question. Our search yielded the research question: “What are the dimensions of leaders’ AI-driven capabilities?” We employed semi-structured interviews to gather primary data from 20 interviewees from 14 financial service companies in Bangladesh, including banks, leasing, insurance, and fintech. We used the online platform Zoom to conduct the interviews. The entire data collection period was four months. The interviews were audio-recorded and later transcribed. The interviews ranged from 32 to 62 min, averaging 44 min. Immediately after each interview, the first author took notes to reflect on the key aspects of AI-driven leadership. After the interview, interviewees were emailed for follow-up questions for further clarification. The interviews started with general questions, then moved toward open-ended questions, and based on the participants’ explanations, progressed to relatively specific questions. While the interviewees elaborated on various capabilities, they particularly stressed on the dynamic nature of those capabilities by focusing on how they were applied to sense and grasp opportunities and re-configure resources. The interviewer remained flexible and open to new knowledge during the interviews. Moreover, various secondary data were collected. These included company annual reports, news features, and other relevant print and digital media.
Sampling and Data Collection
The interviewees were selected from Bangladesh's financial services industry. Compared to other industries, the financial services industry has witnessed widespread adoption of AI (Jahan, 2021). Considering the dynamism and intensity of AI adoption (Khan et al., 2021), we included participants working in executive positions in private commercial banks, leasing, insurance, and fintech companies. We reached the participants through two online alumni association groups. One of the associations consisted of business graduates working in the financial service industry, while the other consisted of executives working in the banking industry. We initially selected six participants (three from each association) for the interview. A snowball sampling method was employed to reach the rest of the 14 participants. The first six participants were requested to select the rest of the interviewees, considering their active working experience and exposure to adopting evolving technologies in the respective organizations. Finally, we found that all the participants actively work with senior management and have experience working on AI-based projects within their respective organizations. The interviewees have diverse job experience ranging from 14 to 22 years. Among the interviewees, 16 worked in banks, while the rest worked in insurance, leasing, housing finance, and fintech companies. Most of the participants have post-graduate business degrees. Seventy per cent of the organizations where interviewees worked with over 2000 employees (see Appendix A). Such interviewee profiles help to gain valuable insights into leaders’ capabilities, particularly in an emerging AI-driven organizational context. After 20 interviews, we concluded the interviews because of data and theoretical saturation (Strauss & Corbin, 1998). Additional interviews were not likely to generate any new knowledge on leaders’ AI-driven capability.
All the interviews were conducted in Bengali language. Back-translation is regarded as a well-known method to ensure equivalence between the translated and original versions (Behling & Law, 2000). Following the back-translation method (Triandis & Brislin, 1984), two independent bilingual translators translated the audio transcripts. The first translator translated the transcripts from Bengali to English, and the second translated the transcripts from English to Bengali. Next, both translated versions were compared for content equivalence and minor revisions were made to develop the final English translation.
Data Analysis
Following Braun and Clarke (2006), we employed thematic analysis to develop open and selective coding for data analysis. To analyze the data, first, we familiarized ourselves with the transcribed and translated interview transcripts. To develop the first-order concepts, we used open coding by rigorously reviewing every twenty lines of the interview transcripts. At this stage, we particularly focused on the vocabulary used by the participants. We read and re-read the transcripts to ensure that all the data set's interesting features are included. Following a methodological approach, we collated the relevant data for each code. Next, we compared the first-order concepts to find the similarities and differences and categorize them into potential themes. We reviewed each theme against the overall pattern of contents in the data set and finalized the second-order themes by reviewing those themes against the codes. Finally, we employed selective coding to classify the second-order themes into aggregate dimensions of leaders’ AI-driven capability. We developed a data structure to show the first-order concepts (leaders’ activities), second-order themes (capabilities), and combined dimensions (capability sets). Figure 1 illustrates the coding process.
Trustworthiness
We ensured trustworthiness standards by following the guidelines in interpretive research that focus on credibility, generalizability, conformability, integrity, and generality (Miles & Huberman, 1994). We ensure credibility by sharing the summary of findings with the participants for their feedback. Moreover, all the researchers provided input in the data collection and analysis phases. We reached adequate participants to achieve data saturation that confirms theoretical generalizability. We followed university research protocol to adhere to the integrity standards. We ensured conformability by conducting adequate interviews and asking the interviewees follow-up questions wherever necessary. To ensure generality, we kept an open mind during the interviews, and the length of the interviews was sufficient to develop in-depth insights into the various aspects of leadership capabilities in the AI environment.
Findings
Dimensions of AI-Driven Capabilities
AI-driven capability refers to the automated technologies that enhance human intelligence. AI-driven capability stimulates reasoning and ultimately results in human-machine collaborative behavior (Sayantini, 2022). AI-driven capabilities include leaders’ capacity to adopt AI technologies, employ data-driven decisions, and coordinate human-machine collaboration in organizations. From a leadership perspective, AI aids in overcoming cognitive limitations and making effective decisions (Jarrahi, 2018). Wilson and Daugherty (2018) also support the arguments that AI can augment leaders’ capability for better decision-making. However, Brynjolfsson and Mitchell (2017) contend that AI competence is effective in specific contexts and is comparatively fragile compared to human decision-making for many tasks. AI-based systems work on data-driven rationality. AI systems often weigh more on value-creating parameters that compromise ethical standards (Wright & Schultz, 2018). In this scenario, leaders’ capabilities with AI capabilities will collaboratively determine the organizational success (Tsai et al., 2022). The interview findings suggest that six dynamic capabilities under three dimensions are vital for leaders in AI environments (see Appendix B). The following sections discuss these dimensions of AI-driven capabilities under each dimension.
Technical Capability
Leaders’ technical capability refers to the resources and infrastructure required for leaders to lead successfully in an AI environment (Davenport & Bean, 2021; Davenport & Foutty, 2021; Davenport & Mittal, 2023; Kolbjørnsrud et al., 2017; Watson et al., 2021). Leaders encounter greater challenges in the absence of technical capabilities. Davenport and Mittal (2023) find that the heads of AI and analytics in most organizations spend significant time highlighting the value and purpose of AI technology to other managers. They focus on educating decision-makers from all organizational units regarding the appropriateness of AI functions and AI-based processes. Apart from upskilling and re-skilling the workforce, organizations must acknowledge that all employees require AI-based training. Table 1 presents the interview findings that suggest 12 technical activities clustered into two technical capabilities.
Technological Capability
Leaders’ capacity to use AI technologies is a key technical capability in AI environments. Technical capabilities aid leaders in enhancing credibility and adapting to changing circumstances (Hysong, 2008). Technological soundness in AI empowers leaders’ cognitive capabilities (Akter et al., 2021b). Leaders’ technological soundness persuades middle and front-line managers to learn new technological skills, establish a data-driven decision-making culture, and drive for innovations in the organization (Davenport & Mittal, 2023). Leaders’ attitudes toward AI-based technologies, the extent to which they are equipped with those technologies, and the relative efficiency of using them are the key elements of technological capabilities (Motamarri et al., 2020).
Most interviewees said that leaders from different organizational levels use AI-based technologies to various degrees. Due to competitive and regulatory pressures, banking organizations are becoming increasingly automated. Leaders in those organizations must use AI-based technologies for internal operations and external regulatory purposes. For internal business operations, the organizations have their ‘core banking software’ which is updated frequently with increasingly automated features. Operational leaders must use such technologies for internal operations and communication purposes. Due to the Bangladeshi government's extended digitalization efforts over the last decade, most government entities have adopted many automated software and online platforms. Leaders from different levels of organization need to report and coordinate with such external entities, which requires them to be capable of using technologies at an adequate level. For example, organizations need to report to the Central Bank, coordinate with the Revenue Board and the Land Management Office (for mortgage purposes), collaborate with law enforcement agencies, etc. ID2, who works with the IT division in a leadership capacity in a private commercial bank, reflected on technological aspects. We acknowledge the need to use AI-based technologies and equip ourselves with sophisticated technological tools. Moreover, top management is increasingly paying interest to invest in AI-based technology infrastructure for strategic purposes. We feel that the capacity of emerging leaders to use such AI-based technologies is and will remain central to the all-out effort to digitally transform the respective organization better than others in the highly competitive financial services industry (ID 2).
AI-based knowledge is crucial to leaders’ technological capability in an AI environment. Adequate knowledge helps to successfully apply AI-based processes and assess the outcomes (Akter et al., 2021b). AI-related knowledge also equips leaders to serve customers better in an AI environment (Motamarri et al., 2020). Most of the interviewees possess knowledge and skills in terms of AI. They also mentioned that some of the Board of Directors (BoD) possess sophisticated AI-related knowledge and keep updates on automated technologies worldwide. However, such knowledge among the executive leaders is significantly higher than that of BoD members. ID18, who works for a large private commercial bank, has active experience working with AI-based projects in the organization. He has experience dealing with BoD and other operational leaders in AI-based projects. As operational leaders, we are required to actively engage and solve many of the routine and unique problems in the work environment. Updated AI-based knowledge enables us to re-think work-related issues relatively easily and cost-effectively. Moreover, strong knowledge is essential to pursue BoD for approval of AI-based proposals. In addition, we consider AI-related knowledge as leverage to their current job security and future job opportunities and competitiveness (ID 18).
Most interviewees said that financial services are getting increasingly digitalized. In an increasingly digitalized environment, AI-based knowledge is an essential skill and a critical parameter of an organization's strategic success.
Informational Capability
AI-based systems and processes generate sufficient information that benefits organizations. Leaders’ capability to access and use AI-based information systems and processes significantly determines the success of AI-driven organizations (Akter et al., 2021b). Such capability includes leaders’ capacity to access AI-based analytics and updates on AI-based systems and processes (Motamarri et al., 2020). In complex circumstances, AI extends leaders’ knowledge (Doornenbal et al., 2022).
The interviewees said that, like ID14, who works as an Assistant Vice president for a private commercial bank, the increasing use of AI-based technologies has opened avenues to generate ample information in many aspects of the organization. Some interviewees, like ID14, were more interested in accessing and relying on automated technology-based information. We feel that access to such information creates the opportunity to decide based on a wide range of data in a relatively unbiased and objective way. Moreover, access to AI-based processes enables us to remain updated on real-time situations in various circumstances. To develop emerging leaders, organizations should create scope to use AI-based information in various situations (ID 14).
Because of the nature of business, leaders also hold a restrictive view of giving access to AI-driven information to all levels of employees. Only the strategic level leaders can access all the AI-based information and processes. The rest of the leaders only have access to AI-based information to the extent they require within their job responsibilities. Otherwise, open access to AI-based information may jeopardize the safety and security stakes of the concerned stakeholders. A well-crafted policy to access the AI-based process and use the AI-based information reduces the risks of stakeholders’ safety and security concerns.
Interviewees said that relevant training on various AI-based information systems and processes aids them in crafting effective strategies in the AI-driven financial services industry. Training is vital in accessing and using AI-related information in various circumstances. Regularly exposed to various training sessions develops leaders’ analytical capacity to deliver better in an AI-driven organizational environment (Motamarri et al., 2020). Most interviewees said they often participate in AI technology-related training, workshops, seminars, etc., at home and abroad. Most organizations have their own training institute. The IT division normally arranges routine technology-related employee training in collaboration with the training institute. Sometimes, unit heads are provided with routine technology-related training and assigned responsibility for training their units. ID1, who works in the human resources department of a private commercial bank, reflected on this perspective. To equip us with cutting-edge AI technology-based skills, the organization undertakes development programs. Senior leaders guide the IT division in designing the development programs. External experts are frequently hired to conduct the development programs. These programs aim to develop emerging leaders. The participants who show higher potential are frequently sent to the development programs at national and international levels (ID1).
Adaptive Capability
Leaders’ capacity to grasp and articulate the effects of AI-based actions is the leaders’ adaptive capability. Adaptive capabilities help leaders to assess situations and choose paths to reach targeted outcomes correctly. Adaptive capabilities also assist leaders in developing and implementing effective production, marketing, and human resource policies for the organization (Campbell, 2021). Decision-making and integration are two key adaptive capabilities in an AI-driven organisational environment. Based on our interview findings, Table 2 shows 14 adaptive activities clustered into two adaptive capabilities.
Data Analysis Process: Technical Capabilities Theme.
Data Analysis Process: Adaptive Capabilities Theme.
Data Analysis Process: Transformational Capabilities Theme.
Decision-Making Capability
Decision-making is considered a leader's significant adaptive capability. Businesses are increasingly making more decisions in an array of areas than ever before. A study by Harvard Kennedy School shows that by 2018, 34% of total jobs require core decision-making in areas like analyzing, prioritizing, and strategizing. AI helps to increase the volume of decision-making in the organization. Often, decision-making in organizations is related to problem-solving. In AI, leaders use analytics insights to decide and solve problems in various situations (Akter et al., 2021b). Problem-solving is related to the leaders’ capacity to fix problems, the tendency to use data over experience, and the exerting creativity in the problem-solving processes (Motamarri et al., 2020).
Most interviewees said that over the last few years, leaders have increasingly relied on automated technologies to solve problems in the work setting. However, many unique problems are difficult to resolve by using AI technologies. Operational leaders focus on the dimensions of those problems and think about how the solutions can be translated into automated technologies. ID5, who leads a private commercial bank's customer relationship management team, explained how to use AI-based technologies to solve organizational problems. We focus on data-driven insights over hunches to solve any problems. We feel that data-based decision making is more scientific and predictable and can be replicated in similar circumstances. It saves time and cost in the long term with greater ease in operations. A focus on creating an AI-driven environment will direct the employees’ thought process toward tech-driven solutions to emerging problems (ID 5).
AI is effective in increasing efficiency and improving accuracy in decision-making (Agrawal et al., 2022). However, improved accuracy and increased efficiency are insufficient because decision-making in one area impacts the other areas of the organization. While designing AI-based decision-making systems, organizational leaders must re-visit the integrated paradigm of the value chain (Agrawal et al., 2022). In the collaborative human-AI decision-making process, Raisch and Fomina (2022) find that probabilistic decision-making is more productive than deterministic decision-making. Following the ‘interactor’ archetype, leaders can effectively decide and solve problems in an AI environment (Meissner & Keding, 2021). Decision-making capabilities include leaders’ capacity to evaluate the impact on jobs, cost, productivity, and timeliness while adopting AI in organizations (Wamba et al., 2017).
Most interviewees said it is important to consider both the short-term and long-term benefits of adopting AI-based technologies against the cost of adoption. To decide whether to adopt AI-based technologies in various circumstances, primarily, they focus on strategic interest. In the short term, they focus on the benefits of saving time and increasing efficiency. ID17, a corporate branch manager of a large commercial bank, reflected on decision-making. In the long run, we assess whether automated technologies will conflict with the overall strategic directions of the organization. Some financial organizations have a strategy to expand toward more rural areas, which restricts over-emphasize automated technologies as it will create a direct negative impact on the prospective client base (rural clients are comparatively less tech-savvy). On the other hand, some organizations are also expanding to the rural areas, but through the agent banking networks which give them higher liberty to use automated technologies to connect their agents (ID17).
Most interviewees also mentioned that the duration of AI-based project implementation and the consequence of such project implementation difficulties on the routine work of employees are two key considerations in deciding on AI adoption. They also consider the impact of automated technology applications on the employment risks of the current workforce.
Integration Capability
Integration capability, which is dynamic (Teece et al., 1997), conceptualizes how leaders coordinate and control AI initiatives in the organization. Coordination involves leaders addressing AI issues in cross-functional meetings and sharing AI-related knowledge within the organization. On the other hand, controlling focuses on leaders’ ideas about AI performance criteria, AI-based business processes and methods, and AI responsibility within the organization. Such capability emanates from leaders evaluating AI-based proposals and using detailed information on AI-based methods and processes (Wamba et al., 2017). Most interviewees emphasized the integration difficulties of AI-driven changes in the organization. They stress arranging a series of cross-functional meetings to discuss the coordination issues. ID11, an experienced leader working for a large private commercial bank, mentioned the unique integration difficulties. If a coordination plan is not clearly crafted and guided with combined efforts before AI- based technology implementation, it ultimately incurs undue cost and time. When any project is undertaken to adopt AI-based technologies in the organization, it is important to assign responsibilities to employees from various hierarchical positions with expected performance standards (ID 11).
Most interviewees emphasized continuously monitoring the progress of the AI-based projects and guiding from time to time. Moreover, integration efforts are important not only for AI applications but also for the changes in associated business models. Interviewees mentioned that organisational objectives remain unfulfilled if AI adoption is not properly aligned with effective business models.
Transformational Capability
The concept of transformational capability is widely used in sociology, politics, and organizational behavior. Leaders’ transformational capability translates employee performances beyond expectations (Khan et al., 2020). In the AI environment, leaders’ transformational capabilities focus on bringing organization-wide AI-based changes (Davenport & Bean, 2021; Davenport & Foutty, 2021; Watson et al., 2021). Systematic AI applications in various organization functions enhance data-driven decision-making and new business processes (Davenport & Mittal, 2023). Likewise, AI must promote products, service offerings, and new business models. Based on our interviews, Table 3 presents 10 transformational activities clustered into two transformational capabilities.
By nature, digital leaders need to be transformative. However, transformational leadership scales require re-conceptualizations in AI-driven dynamic environments. Because transformative leadership competencies are vital for an organization's digital transformation (Schiuma et al., 2024), insights of leaders’ AI-driven capabilities will implicate transformational leadership scales, ultimately leading to successful digital leadership in an increasingly AI-driven environment.
Sense-Making Capability
Sustainable AI-driven transformations depend on effectively developing an organization's sense of AI initiatives. Sense-making is a critical transformational leadership capability (Lugtu, 2020). In an AI-driven organizational context, sense-making includes leaders’ ability to grasp AI initiatives’ near and distant future (Rafferty & Griffin, 2004). The interviewees have diverse views in terms of short- and long-term objectives of AI-based technology adoptions. Through various seminars, symposiums, workshops, etc., the tech-savvy leaders always convey that the organization is heading toward digital transformation in the long run. They also try to keep employees’ confidence by assuring them that their jobs wouldn’t be jeopardized even at a level of advanced automation. They encourage employees to develop skills in the same direction. On the other hand, operational leaders view differently in terms of sense-making. ID12, who has a decade of experience as a leader with a private commercial bank, mentioned the short-term aspects of AI adoptions. We are more interested in guiding employees in the short run. We always focus on continuous learning, and guide employees to learn new technologies. We constantly communicate employees regarding the possible changes of AI-based technologies in the short run. As AI-based technologies rapidly change, understanding short-run implications is critical for employees (ID 12).
Often, organizations offer new business models associated with AI-based changes. Lugtu (2020) refers to rethinking old organizational issues in novel ways that lead to model innovation. The new model stems from how leaders re-think old organizational issues, question the existing organizational practices, and challenge the underlying assumption of such practices (Rafferty & Griffin, 2004). Most interviewees mentioned that senior leaders’ initiation of new business models is critical to making sense of AI-based technology implementation. However, AI-based projects do not always require a business model. Still, adopting automated technologies impacts the existing target markets by giving the same services in better ways. Sometimes, such technologies primarily focus on distinct client bases (internet banking, apps, etc., focus on professionally elite groups of clients). ID11, who has extensive experience and works in the top management of a private commercial bank, reflected on the relevance of new models. We want to re-think the old organizational issues in automated technology-based ways. We focus on the assumptions of organizational existing working mechanisms. If changes in the assumptions are possible by using AI-based technologies, the outcomes would be more fundamental and sustainable. This will significantly impact the organization's competitive edge (ID11).
Uncertainty-Dealing Capability
Teece et al. (2016) introduce uncertainty-dealing capability as a transformational leadership dimension. In an increasingly AI-driven environment, presuming the unpredictability of AI-based changes on different work units, understanding the consequential severity of AI-driven changes, and relying on AI-based technologies to address those consequences are the key elements of leaders’ uncertainty-dealing capability (Matsunaga, 2022). Most of the interviewees said that it is important to recognize both the internal and external uncertainties arising out of automated technology applications, such as whether employees will be able to adopt such technologies, whether these will impact work teams positively, and whether such applications will bring any unprecedented disruptions in the internal working environment. Regarding external implications, leaders are more concerned with the uncertainty regarding whether the AI-aligned business model will work properly and whether the existing client base will be misguided due to the automated technology applications. ID15, a highly experienced top-level leader working for a leasing company, illuminated the AI-driven uncertainty aspects at the workplace. So, as a part of top management, we need to project the uncertainties while adopting AI technologies in the organization. Moreover, we must assume that uncertainties are normal when adopting such evolving technologies. It is more important to stay confident and be ready to handle any uncertainties by creatively using AI-based technologies (ID15).
Most interviewees mentioned that tech-savvy top leaders are more confident and capable of dealing with uncertainties. The rest of the leaders rely on the IT division's proactiveness to properly assess the consequences of AI-based technology adoptions.
Discussion
Our research findings on AI-driven capabilities extend our understanding of the DMC view. Dynamic managerial capabilities enable organizations to create, expand, and modify their strategies and fulfil their mission. Leaders can use dynamic capabilities to influence the external environment and impact the internal attributes of the organization (Harris & Helfat, 2016). They use dynamic capabilities to develop and deploy organizational resources and capabilities to sustain successful organizational outcomes. Leaders from both top and mid-levels may possess dynamic managerial capabilities. The findings offer insights into the three dynamic managerial capabilities: sensing, seizing and reconfiguring (Appendix B). Next, we will explain how our findings on technological and informational (technical dimension), decision-making and integration (adaptive dimension), and sense-making and uncertainty-dealing (transformative dimension) leader capabilities inform each of the three dynamic managerial capabilities: sensing, seizing and reconfiguring.
Our interview findings suggest that technical capability, consisting of technological and informational capabilities, significantly enhances leaders’ sensing capacity in an increasingly evolving technological landscape. The sensing capability refers to the leaders’ capacity to identify customers’ needs and preferences by observing the organizational environment and developing the latest technologies (Engelmann, 2023; Teece, 2012). However, leaders must use AI and thoughtfully interpret the outputs based on data-driven insights to apply sensing capability effectively. Successful organizational leaders use various AI technologies and technology-driven information to uncover emerging opportunities in the external environment (Leachman & Scheibenreif, 2023). For example, leaders’
The adaptive capability consisting of decision-making and integration capabilities promotes organizational seizing capacity in the competitive environment. Adaptive capability leads to seizing new market opportunities (Dhar et al., 2024). The seizing capability refers to the leaders’ capacity to make decisions on strategic and business models to create value for the customers and organization (Helfat et al., 2007). However, effective seizing capability requires leaders to make appropriate and timely decisions on adopting AI technologies and successfully integrate AI-based strategies in organizations. Organizational leaders recognize the importance of adopting emerging AI and machine learning technologies (Leachman & Scheibenreif, 2023). However, our interview findings show that adaptive capability reflects decision-making and integration capabilities that enable leaders to seize opportunities successfully. For example, leaders’
The interview findings show that transformational capability facilitates organizational reconfiguring capacity to meet the changing demands of the external environment. The transformational capability is reflected in leaders’ sense-making and uncertainty-dealing capabilities. The reconfiguring capability refers to the leaders’ capacity to bring strategic renewal of the organization's resources and capabilities to meet the changing demands of the external environment (Engelmann, 2023; Helfat & Campo-Rembado, 2016). However, successful reconfigurations require organizational leaders to make sense of AI-based changes and deal with the uncertainties arising out of such changes. Reconfiguration involves changing various aspects within the organization, and sense-making creates a shared understanding of the changes among the employees (Aoki, 2022). Leaders’
Theoretical Contributions
There are several theoretical contributions from this study. First, through the DMC lens, this research provides a novel framework on leadership capabilities to lead AI-driven organizations (Appendix B). We contribute to the DMC view by studying DMC in an AI-driven organizational context. Individual-level dynamic capabilities are required to anticipate and interpret disruptive changes in the external environment (Helfat & Peteraf, 2015). In the AI environment, contemporary capability research is conceptualized from a service analytics perspective (Akter et al., 2021a; Motamarri et al., 2020; Wamba et al., 2017). This research shows that leaders must re-think their capabilities to lead AI-driven organizations more transformatively. Leaders’ AI-driven capabilities will significantly influence other areas beyond service analytics. AI-driven capability enables leaders to re-design, prototype, and offer personalized products that accelerate development, saving time and money. AI-driven leaders can also forecast demand better and optimize logistics for efficient supply chain management. Moreover, AI-driven leaders can streamline the organizational recruitment system and ensure better employee engagement by analyzing survey responses and communication patterns that enhance the organization's human resource and talent management system. To do these, leaders will require human intelligence capabilities like AI-based knowledge, creativity, and judgment to augment the AI capability. Collaborative human-machine capabilities will increasingly determine the best strategic directions for competitive organizations (Davenport, 2016). The combination of human-machine capability creates the outcomes from DMC in AI environments.
Second, we set out to identify the AI-driven capabilities of executives from the fast-developing Bangladeshi financial industry. In analyzing the findings of those AI-driven capabilities required in an AI-driven context, we found that these AI-driven leader capabilities map well with transformational leadership. The leaders’ capability to transform their followers through the digital transformation of the business environment is increasingly becoming apparent (Li et al., 2016). We take the first steps towards this line of inquiry by highlighting transformation capability as a must-have capability for leaders in increasingly AI-driven organizational environments. However, our study findings do not go as far as claiming that the AI-driven capabilities of leaders can “increase” transformational leadership qualities. We argue that AI-given capabilities and transformation leadership attributes are compatible. Organizations can operationalize AI-driven capabilities through key transformative leader attributes for successful digital leadership. In AI adoption, transformational leaders are better positioned than other types of leaders to foster employee through creativity, adaptability, and innovation to achieve organizational agility (Gyanamurthy & Radhanath, 2023); to address employee concerns about job security and workforce changes by fostering a positive perspective on AI's benefits (Kravchenko, 2019); understand the impact of AI on the workforce and ensure that each individual's role in the transformation is valued (Nicastro, 2023); inspire employees to overcome fears, providing tailored support to ease the transition (Hoch et al., 2018; Matsunaga, 2022) and effectively communicate the vision and benefits of AI, to facilitate a smoother adoption process. We leave it for future studies to investigate the validity of our claim and the extent of alignment/congruence between AI-driven capabilities and transformational leadership attributes.
Third, we also advance the current stream of AI-driven leadership research (Davenport & Bean, 2021; Watson et al., 2021) by offering an AI-driven capability from a DMC view. We propose novel AI-driven capabilities. Previous DMC research focused on uncertainty and crises (Parker & Ameen, 2018). Leaders can leverage AI to handle uncertain and crises. By predictive modeling, leaders can assess the impact of crises while a vast amount of data aids in detecting the early signs of a potential crisis. Moreover, leaders can use AI-driven training to develop an adaptive learning system in the organization that creates employee resilience during uncertain and crises. Additionally, the successful use of automated communication tools enables leaders to efficiently communicate with concerned stakeholders during crises. Transformative AI can play a vital role within the organizational business process to equitably accommodate the needs of all relevant stakeholders (Kompella, 2022; Marr, 2020).
Finally, this research contributes to the Upper Echelon Theory (UET) in several ways. UET conceptual model argues that top management decision makers’ cognitive frame reflects strategic decisions in the organization (Hambrick & Mason, 198419847]). Cognitive heterogeneity enhances leaders’ decision-making capabilities based on higher information processing capabilities, ultimately impacting organizational performance. However, how such capabilities are framed in a technology-driven, particularly AI-driven organizational environment remains largely underexplored (Krakowski et al., 2022). Our interview findings offer collaborative human-machine capabilities under three dimensions that impact upper-echelon leaders’ cognitive frame to make strategic choices. Data-driven insights and predictive capabilities enable leaders to make enhanced decisions, while personalized learning and bias reduction enable them to gain greater managerial efficiency. Moreover, AI-driven capabilities can promote new business models and foster organizational strategic innovations.
Practical Contributions
Our research findings contribute to the increasingly AI-driven organizations in three ways. First, understanding DMC in an AI-driven organizational context offers leaders skills to face disruption and navigate changes effectively to maintain competitive advantage in a rapidly evolving environment (Heubeck, 2023). Leaders with DMCs can effectively sense, seize and reconfigure according to the changing preferences of customers, investors, suppliers and regulatory bodies. This leads to better stakeholder relationship management. Moreover, DMCs will allow leaders to swiftly align organizational resources with strategies that help to create organizational agility. Organizational agility is vital for survival and growth in highly volatile and rapidly evolving industries (Eilers et al., 2022). Second, we examine six AI-driven capabilities. Increasingly, AI-driven organizations may consider developing those capabilities among emerging leaders through structured training and development programs. The respective organizations will also be able to identify the more promising, responsible, and transformative leaders suitable for AI environments. Building digital capabilities among the workforce will benefit organizations in competitive environments (Yokoi et al., 2023). Third, AI adoptions pose ethical risks for organizations and create new leadership challenges. We argue that leaders’ AI-driven capabilities are vital in addressing those challenges. Hence, developing leaders’ AI-driven capabilities will eventually equip them with the knowledge of AI ethics. Ethics programs in organizations should start from the executives, then go to the ranks, and finally be translated into AI technologies. If leaders possess knowledge of AI ethics, translating the ethical considerations will become effective and sustainable. Leaders’ ethical awareness mitigates the transparency and discrimination of AI algorithms and better customer privacy and service personalization (Yokoi et al., 2023). AI ethics helps organizations adopt value-driven technologies, enact better policy structures, and set exemplary corporate standards (Yao et al., 2018).
Limitations and Future Research Directions
Our interview findings provide novel and significant research directions for AI-driven leadership. However, inherent to interview-based studies, our research also has several limitations. First, the scope of our research is limited to 20 interviews in the financial services industry, and other contexts may explore distinct dimensions of capability. Although the interviews provide deep insights based on individual experience, the broader context may be overlooked, limiting the generalizability. Future research may investigate other contexts based on a greater sample size to build a general understanding of leaders’ AI-driven capability. Second, we don’t consider diverse contextual factors such as the level of AI adoption or firm size to explore the dimensions of AI-driven capability. Future research may empirically consider the firm-specific factors to overcome such limitations. Third, we explore only the three dimensions of leaders’ AI-driven capability. Multiple cross-sectional and longitudinal studies may investigate our research findings empirically. Cross-sectional studies will help to understand the contextual factors, while longitudinal studies will aid in rigorously comprehending the AI-driven capability as a dynamic capability. Fourth, the interviewees in this research participated as individuals rather than organizational representatives. This limits participants’ capacity to disclose confidential but contextually relevant issues to the AI adoption. Future research may focus on selecting participants subject to the approval of respective organizations that will aid in capturing more organizational issues related to AI.
Finally, we extend the DMC discourse within a complex AI-driven organization system. We synthesize the extant literature with our interview findings on leaders’ AI-driven capabilities. The dynamic capability is influenced by various inhibiting and enabling managerial factors (Ambrosini & Bowman, 2009). AI-driven organizations offer new products, services and business models in the changing competitive landscape. However, organizational leaders must be more agile to bring innovations in the fast-changing business environment (Eilers et al., 2022). Future DMC research may focus on uncovering how leaders can use the DMCs to be more agile and innovate in disruptive environments to sustain competitive advantage.
Conclusion
Digital leaders are called upon to successfully manage the digital transformation and the organization in an increasingly AI-driven environment. Combining AI and human capabilities, we explore the dimensions of collective intelligence in leadership and current and future AI applications. The extant management literature acknowledges the critical role of AI research in leadership. Leaders must fundamentally re-think how humans and machines interact at work (Davenport & Mittal, 2023). We share a novel AI-driven digital leadership capability framework to lead AI-driven organizations effectively (Appendix B). Although embryonic, it shows how AI influences the current leadership paradigm, requiring new conceptualizations to advance the theories in emerging management streams of knowledge.
Footnotes
Declaration of Conflicting Interests
Funding
ORCID iDs
