The AI landscape

Despite the extensive interest in the benefits of AI, relatively little attention is shown to the dark side of AI. Senior organizational leaders must prepare accordingly to address the challenges emerging from AI’s dark side. A Governance Institute of Australia survey shows that 94% of respondents claim to involve board members in technology and cyber issues (Tong, Reference Tong2022). However, incidents such as the 2022 Optus scandal in Australia clearly show that the board members were not equipped to address the risks that emerge with cyber security threats and constantly evolving technology (Tong, Reference Tong2022). Senior organizational leaders need digital literacy similar to when they were expected to develop financial literacy after Enron and other scandals after 2001.

AI can potentially produce the risks associated with individual, organizational, and societal levels, which are significant elements of digitalization (Alt, Reference Alt2018). At the individual level, the negative effect of AI is reflected in privacy concerns and product and content recommendations. AI gains deep insights into privacy concerns (Grewal, Guha, Satornino & Schweiger, Reference Grewal, Guha, Satornino and Schweiger2021). For example, voice assistants like Alexa could predict key moments by analyzing the customers’ voices with AI technology. Moreover, facial recognition-based payments exacerbate privacy risks as the human face is used as a proxy for various personal information, including age, gender, and appearance. Li, Zhao, Hussain, Ming and Wu (Reference Li, Zhao, Hussain, Ming and Wu2021) find that personalized recommendations narrow perceived information and privacy concerns, resulting in a reluctance to accept technologies.

From an organizational standpoint, AI-enabled products are likely to impact the profitability and reputation of the companies. For instance, if the AI-enabled chatbots do not deliver as expected, customers will distrust them, creating a trust gap between the customers and companies (Yen & Chiang, Reference Yen and Chiang2021). Moreover, organizations also face enormous challenges in implementing AI strategies and often fail to assess the impact of AI technologies on the human workforce (Cheng, Lin, Shen, Zarifis & Mou, Reference Cheng, Lin, Shen, Zarifis and Mou2022). From a societal perspective, AI yields dark effects ranging from workforce replacement to ethical problems. Danaher (Reference Danaher2019) finds that AI brings widespread concerns for the human workforce. Beyond e-commerce, AI exists in most spheres of the daily lives of many and is likely to increase unemployment. Moreover, issues like AI regulation, moral dilemmas, AI fairness, and discrimination are increasingly posing challenges to AI governance at societal levels (Wirtz, Weyerer & Sturm, Reference Wirtz, Weyerer and Sturm2020).

Concerns in AI and leadership

Due to cloud-based data processing, challenges in geography, technology, and policy issues have emerged. Combined with AI, the Internet of Things, analytics, and virtual and augmented reality affect many business strategies and operations (Davenport, Reference Davenport2019). Among such key emerging situations, ethical concerns about AI use by organizations are rising. AI is emerging with more ethical risks because of probability-based AI decisions, the continuously changing external environment, and AI’s inherent algorithmic complexities (Qi, Sun & Hosseini, Reference Qi, Sun and Hosseini2023). Organizational leaders can’t remain indifferent to AI-driven ethical concerns because these will impact employees, customers, suppliers, brands and reputations, and broader stakeholders (Davenport & Katyal, Reference Davenport and Katyal2018). For example, the Optus security breach in Australia showed a lack of responsibility by senior leadership. A survey conducted by Governance Institute Australia shows that 34% of respondents reflect on their board members’ capability to deal with such situations. Around 50% of respondents view the Optus security breach occurred because of data management policy failure, which is the responsibility of senior leadership (Tong, Reference Tong2022). Responsible leaders can successfully address ethical concerns stemming from accountability, fairness, transparency, and safety issues in the AI landscape.

Accountability in the AI context focuses on human liability over AI’s design, implementation, and monitoring phases (Leslie, Reference Leslie2019). Leslie (Reference Leslie2019) views that AI cannot justify its decisions and requires human judgment to assess the effects on various stakeholders. Leading responsibly focuses on human control over AI lifecycle stages by recognizing responsibility (Kompella, Reference Kompella2022), answerability, and auditability (Leslie, Reference Leslie2019) at each stage. Fairness refers to AI’s unbiased and justified application for all the concerned segments (Kompella, Reference Kompella2022). This emerges as a key AI ethics dimension because human beings are limited by their contexts and biases when designing AI (Leslie, Reference Leslie2019). Leslie (Reference Leslie2019) also argues that human misjudgment and prejudice create biases in data extraction, processing, problem formulation, and model development stages of AI-based solutions. Leading responsibly in AI environments safeguards fairness through equitable analytical structures, reasonable features, and processes of an AI-based model (Leslie, Reference Leslie2019).

Kompella (Reference Kompella2022) mentions transparency as an important dimension of AI ethics that ensures trust in AI-based decisions. Transparency clarifies the contents of AI systems by answering the question ‘Why’ concerning the AI process. Transparency produces ethically permissible, non-discriminatory, and publicly trustworthy AI-based solutions (Leslie, Reference Leslie2019). Responsible leadership promotes transparency to explain to the affected stakeholders why and how the AI model works in a specific context (Leslie, Reference Leslie2019). Such explanation ensures rationality behind AI-based decisions by clarifying the contents of AI models. From an AI ethics perspective, transparency justifies outcomes of both AI-based model design and implementation processes (Leslie, Reference Leslie2019). Safety, as a dimension of AI ethics, ensures control over the AI system (Kompella, Reference Kompella2022). Leslie (Reference Leslie2019) argues that the technical sustainability of AI systems depends on safety, security, robustness, reliability, and accuracy. Responsible leaders evaluate the transformative effect of the features of technical sustainability on individuals and society in the long run (Leslie, Reference Leslie2019). Producing safe and reliable AI-based solutions is critical from an ethical perspective. Potential failures of AI systems may create harmful impacts and ultimately damage public trust in AI-based solutions. Building safe AI is challenging, especially in volatile, flux, and uncertain contexts. An AI system is safe when it operates dependably and accurately to the designer’s expectations, even in changing circumstances (Leslie, Reference Leslie2019).

Needs for leaders’ AI-driven capability

Extant research on requisite capabilities in AI-driven leadership presents fragmented findings that are inadequate to answer the RQ. We review selected high-quality review papers to present the gap in extant research on leaders’ AI-driven capabilities (see Table 1). For example, Kemp (Reference Kemp2024) highlights conceptually the critical role of AI technologies in gaining a competitive advantage. This paper focuses on organizational activities like grounding, bounding, and recasting and introduces the situated AI concept to achieve competitive advantage. The paper argues that situating organizational activities appropriately enables organizations to be strategically cost-effective in the competitive environment. While the paper recognizes the role of reconfiguring the organizational activities, the individual capabilities required to re-arrange such activities remain beyond the scope of the paper. Tsai et al. (Reference Tsai, Marshall, Choudhury, Serban, Hou, Jung and Yammarino2022) review human–robot collaboration research from various disciplinary perspectives: engineering, psychology, management, and economics. They found that engineering represents the highest 58% of extant research on human–robot collaboration. The psychology discipline presents 34%, while the management and economics disciplines present only 6% and 2%, consecutively. However, with the greater presence of automation in the workplace, management and economics researchers are increasingly investigating various aspects of human–machine collaborations (Tsai et al., Reference Tsai, Marshall, Choudhury, Serban, Hou, Jung and Yammarino2022).

Table 1.

Extant research on leaders’ AI-driven capability

Krakowski et al. (Reference Krakowski, Luger and Raisch2022) focus on effective mechanisms to adopt AI in the organization. They argue that effective AI adoption can improve performance and drive the organization to achieve strategic competitiveness. The paper presents how AI can simultaneously substitute and complement human tasks and rationalize organizational decision-making. From a resource-based view, the paper argues that AI generates substitution–complementation dynamics in the organization. Such dynamics result in suggesting new human–machine capabilities over the traditional managerial capabilities. The paper concludes by highlighting organizations must identify and develop those novel sets of capabilities to stay relevant in the emerging competitive landscape. Banks et al. (Reference Banks, Dionne, Mast and Sayama2022) reconceptualize leadership in the emerging digital era. Studying leadership in informal contexts (e.g., social media), they highlight the need for digital leadership to deal effectively in virtual contexts and in computational modeling (e.g., big data and ML). The paper conceptually addresses the 5Ws (e.g., who, what, when, where, and why) aspects of digital leadership and concludes that practitioners and scholars should reconceptualize leadership in the changing digital landscape.

Doornenbal, Spisak and Van der Laken (Reference Doornenbal, Spisak and Van der Laken2022) focus on the role of ML to understand better the complex pattern of leadership behavior. The paper argues that the trait leadership approach seems to be more suitable for increasing automation in organizations. They find that ML can effectively assess complex relationship patterns by interpreting algorithms’ outcomes. Using a Big Five inventory framework of cognition, they quantitatively assess and demonstrate the outcome of leadership behaviors. The paper concludes that the trait approach of leadership aligns better to predict the behavioral model complexity and generate interpretable results. Extant research reviews show that new forms of leadership, such as digital leadership and AI-driven leadership, are increasingly becoming relevant in the emerging digital landscape. However, the focus on the required capabilities to execute the exhibited leadership approaches is inadequately examined in the extant research. Drawing on these findings, this paper integrates the relevant literature on the management and marketing domains to uncover the dimensions of leaders’ required capabilities in the AI environment. Following the research gap, we discuss the technical, adaptive, and transformational dimensions of leaders’ AI-driven capability.

The DMC view

Over the last two decades, an emerging body of literature has sought to integrate human psychology insights to clarify understanding of the changing competitive industry structure (Peteraf & Shanley, Reference Peteraf and Shanley1997) and the nature of cognitive biases in strategic decisions (Bateman & Zeithaml, Reference Bateman and Zeithaml1989). During a period of change in the competitive market, an organization’s outlook shifts away from the external environment to internal resources and capabilities (Hodgkinson & Healey, Reference Hodgkinson and Healey2011). Strategy scholars have been paying increasing attention to the behavioral and cognitive processes, highlighting the capabilities to promote organizational adaptation, learning, and performance (Helfat et al., Reference Helfat, Finkelstein, Mitchell, Peteraf, Singh, Teece and Winter2007). The DMC is an effective theoretical lens in capability research from a micro-foundation perspective (Helfat & Peteraf, Reference Helfat and Peteraf2015). Micro-foundations are context-specific (Felin & Powell, Reference Felin and Powell2016); therefore, reviewing required capabilities in AI-driven contexts would call for more effective leadership in the less understood AI environmental context (Felin, Foss, Heimeriks & Madsen, Reference Felin, Foss, Heimeriks and Madsen2012). Understanding AI-driven capabilities as DMCs also allows leaders to decide the strategic factors and set priorities for higher organizational performances (Helfat & Martin, Reference Helfat and Martin2015).

DMCs are the mechanisms that use organizational resources to harness opportunities and neutralize threats in the external environment (Sirmon, Hitt, Ireland & Gilbert, Reference Sirmon, Hitt, Ireland and Gilbert2011). The increasing adoption of AI in organizations is coupled with unprecedented emerging concerns (Davenport, Reference Davenport2019), creating distinctive leadership challenges (Kompella, Reference Kompella2022). DMCs can play a significant role in responding to such leadership challenges. Ishida (Reference Ishida2020) finds DMCs critical in responding to crises, emergencies, and disruptive circumstances. Because of AI’s inherent algorithmic complexities and rapidly changing external environment, leaders must address unique challenges swiftly and responsibly (Babic, Cohen, Evgeniou & Gerke, Reference Babic, Cohen, Evgeniou and Gerke2021). Managing effectively, digital responsibility enables leaders to extend the benefits and offset the risks in a competitive environment (Yokoi et al., Reference Yokoi, Goutas, Wade, Zahn and Niniane2023). Such leadership practices enhance digital performance and aid in achieving organizational objectives in the long run.

Adner and Helfat (Reference Adner and Helfat2003) emphasized the role of managerial capabilities in creating, integrating, and reconfiguring organizational resources within the dynamic capability framework. DMCs arise from three key elements: human capital, social capital, and cognition (Helfat & Martin, Reference Helfat and Martin2015). These elements, independently and in combination, shape an organization’s ability to sense opportunities, seize them, and reconfigure resources accordingly. Human capital is built through managerial knowledge, skills, and expertise, contributing to varying performance levels in similar challenging situations. A diverse and complementary human capital base enhances organizational performance (Wright, Coff & Moliterno, Reference Wright, Coff and Moliterno2014). Social capital refers to managers’ effective combination of resources, enabling improved performance. Leaders leverage their networks and relationships to access specialized knowledge supporting business objectives (Ambrosini & Altintas, Reference Ambrosini and Altintas2019). Cognition, which is linked to personal beliefs and knowledge, involves the mental processing of information. Since human thought influences actions and behaviors, managerial cognition can be studied and analyzed (Taylor, Reference Taylor2005).

Planning and searching

We carefully explored various newspaper articles, industry reports, magazines, and scholarly databases to devise an original RQ. This search yielded the RQ: ‘What are the dimensions of leaders’ AI-driven capabilities?’ To address the research question, this research follows the guidelines by Sheng, Amankwah-Amoah, Khan and Wang (Reference Sheng, Amankwah-Amoah, Khan and Wang2020) to review peer-reviewed empirical quantitative and qualitative articles and review articles and conference papers published in English within the last 10 years in business and social sciences. We focused on the last 10 years because of the significant technological disruptions that changed our way of life, work, and socialization (Balachandran, Reference Balachandran2019). We avoided including non-English sources in our study scope because of time and resource commitment barriers (Rockliffe, Reference Rockliffe2022). Moreover, we tried to ensure a uniform data extraction and synthesis approach. Additionally, this paper is conceptual in nature and doesn’t focus on any specific geographical region. Therefore, we strictly included the papers published in English. Our RQ is founded on the emerging ethical problems in business organizations. Therefore, we intentionally limited our search scope to the business and social sciences and avoided other interdisciplinary fields. We also excluded non-peer-reviewed publications and unpublished studies from our study scope to ensure rigor and avoid bias in our findings (Egger, Juni, Bartlett, Holenstein & Sterne, Reference Egger, Juni, Bartlett, Holenstein and Sterne2003). To conduct this review, keywords, titles, and abstracts of articles are searched in Emerald Insight, EBSCOhost Business Source Complete, and ScienceDirect databases. To extract the documents, the keywords: (‘Artificial intelligence and leadership’ OR ‘AI and leadership’ OR ‘AI and management’ OR ‘AI-driven leadership’) AND (‘Machine learning and leadership’ OR ‘ML and leadership’ OR ‘Machine learning and management’) AND (AI-driven capability* OR AI-driven leadership capability* OR AI-driven managerial capability*) have been searched. Excluding the duplicates, the first round of searching these databases generated 1163 records.

Screening and extraction

In the screening phase, after adjusting for language, year, research field and type of documents, 153 records were identified. Following the guidelines articulated by Akter et al. (Reference Akter, Motamarri, Sajib, Bandara, Tarba and Vrontis2021a), this research rigorously reviewed the titles and abstracts of these 153 records to determine their relevance to the purpose and scope of this research. This review process screened out 124 items and the rest of the 29 papers were selected for full-text review. To examine the eligibility, the abstract searches focused on whether the papers have discussed leadership or managerial capabilities in the AI environment perspectives. Papers that particularly discussed AI from organizational outcome perspectives were excluded from this research. The cross-citation check of the full-text review of 29 papers resulted in incorporating eight more papers in the final sample relevant to this research’s eligibility criteria. Finally, we reviewed those 37 papers to address the RQ. All the articles were published in business, management, and marketing. Nine articles were selected from the MIT Sloan Management Review, seven from the Harvard Business Review, and five from the Academy of Management Review. The 37 papers were selected from a total of 12 journal outlets. All the journals rank Q1 in the Scimago ranking. Of these journals, 12 articles were selected from ABDC A* ranked journals, 22 articles from A-ranked journals, and the remaining 3 from B-ranked journals. Regarding ABS ranking, 7 articles were selected from ABS 4* ranked journals, 3 from ABS 4 ranked journals, 23 from ABS 3 ranked journals, and the remaining 3 articles from ABS 2 ranked journals (see Table 2). In its embryonic stage, AI emerged as a popular area of research where substantial systematic reviews were conducted. Hence, we excluded the papers appearing in ABS-1 and ABDC-C journals to maintain rigor and avoid flawed findings or speculative theories. Therefore, some journals that publish AI papers are omitted from this review. The review results show that most papers focus on AI to link or extend current theories (N = 29; 78%). Some papers review and summarize existing theories (N = 5; 15%) to consolidate knowledge, evaluate progress and provide future research directions. However, a few, contradict current theories (N = 3; 7%).

Table 2.

List of journals included in the sample

* highest ranking.

Synthesis and theme identification

This section presents findings on dimensions of AI-driven capabilities based on papers reviewed for thematic analysis to enact a conceptual model. The model clarifies the antecedents of AI-driven capability from a leadership outcome perspective. This research rigorously studied 37 papers to generate prospective themes. Thematic analysis identifies, analyzes, and reports patterns within the extracted literature (Braun & Clarke, Reference Braun and Clarke2006). In this phase, to explicate the relevant themes, this research utilizes a coding-based analysis technique (Miles & Huberman, Reference Miles and Huberman1994). As mentioned earlier, there is a proliferation of dubious findings on AI research appearing in less quality rated outlets. We devoted sufficient focus to theoretical issues to maintain the rigor of our research by limiting the coding to a certain standard of Scimago, ABS, and ABDC-indexed papers. Initially, two authors coded each paper which achieved a consensus of 81% (30 papers). They reconciled the contrasting opinions until a 100% consensus was achieved. The third author facilitated the discussion to reconcile the contrasting opinions between the initial two authors (Dadich, Abbott, Lux & Lowe, Reference Dadich, Abbott, Lux and Lowe2024). This process explored three themes that form AI-driven capability: technical, adaptive, and transformational capabilities (see Figure A1). In the first phase of this process, we repeatedly read 37 articles initially selected from the systematic review process about leaders’ AI-driven capability. In the second phase, we developed codes based on inherent data from the reviewed articles. We developed 10 codes for technology, information, knowledge, training and development, problem-solving, decision-making, sensemaking, model innovation, integration and reconfiguration, and uncertainty dealing. In the third phase, we uncovered the thematic gaps and identified three themes: technical capability, adaptive capability, and transformational capability. In the fourth phase, we report the details of the findings and propose an integrative conceptual framework.

Furthermore, we conducted a quality appraisal of the reviewed papers in this SLR. Quality assessment is paramount in SLR because the quality of the conclusion depends on the quality of the selected literature (Yang et al., Reference Yang, Zhang, Shen, Huang, Zhou, Rong and Shao2021). We followed four guidelines to assess the quality of the reviewed papers (UNC, 2025). First, we examined each paper’s relevance to this study by evaluating their appropriateness regarding our RQ. We also evaluated whether the selected papers could be aligned with our study scope. Second, the reviewed papers were scrutinized to explore reliability. At this stage, we ensured each paper was included in multiple databases. Additionally, we checked whether the paper’s report method was clear. Third, we ensured the validity of the included papers by evaluating their findings based on the claims mentioned in the RQs and methods. We also checked the limitations and potential for generalizability of findings to examine the strengths of those papers. Fourth, we ensured the applicability of the included papers by evaluating those in terms of our study context. We examined whether the papers discussed AI from managerial, leadership, and organizational capability development perspectives to ensure that the included papers fit better with our study context.

Technical capability

Leaders’ technical capability (Davenport & Bean, Reference Davenport and Bean2021; Davenport & Foutty, Reference Davenport and Foutty2021; Davenport & Mittal, Reference Davenport and Mittal2023; Kolbjørnsrud et al., Reference Kolbjørnsrud, Amico and Thomas2017; Watson et al., Reference Watson, Desouza, Ribiere and Lindič2021) refers to the infrastructure and resources that make a leader capable of successfully leading in AI-driven organizations. In the context of AI-driven organizations, leaders with technical capability can understand, use and guide various aspects of automated technologies to their teams, projects, and overall organizations. Such capability emerges from individual leaders’ domain knowledge, expertise, and experiences in respective technologies (Watson et al., Reference Watson, Desouza, Ribiere and Lindič2021). For example, a chief executive officer, who understands the automated technologies better, can make more informed decisions about investment in AI technologies for the organization. Evolving technology-related knowledge, understanding of standards, industry trends, real-world technological problems, and proficiency in using automated technologies indicate leaders’ technical capability in an AI environment. Davenport and Mittal (Reference Davenport and Mittal2023) note that leaders face greater challenges when they lack technical capability. They also find the heads of AI and analytics in most organizations spend significant time with other managers to highlight the purpose and value of AI technology. They highlight educating decision-makers from all business units regarding the process and appropriateness of AI functions. Still, most organizations require upskilling and reskilling their workforce and changing the notion that all employees do not require AI training (Davenport & Mittal, Reference Davenport and Mittal2023).

Leaders’ capabilities to use AI-based technologies, access AI-related information, deploy AI-based knowledge, and transform the learning from AI training into actions are described as technical capabilities in the literature. Technical capabilities not only act as a source of leaders’ credibility but also aid leaders in adapting to changing contexts (Hysong, Reference Hysong2008). Leaders’ technological soundness is a significant dimension of empowering cognitive capabilities in AI-driven organizations (Akter et al., Reference Akter, Wamba, Mariani and Hani2021b). Senior leaders’ interest in technology is particularly important in persuading middle- and frontline employees to adopt new technological skills, drive innovations, and establish a data-driven decision-making culture (Davenport & Mittal, Reference Davenport and Mittal2023). Leaders’ attitudes toward technologies, relative dexterity in using such technologies, and the extent to which the leader is equipped with those technologies have been emphasized in the literature (Motamarri et al., Reference Motamarri, Akter and Yanamandram2020). The leaders’ access to information has been identified as a key capability to perform in an AI-driven environment (Akter et al., Reference Akter, Wamba, Mariani and Hani2021b). This capability includes leaders’ access to AI-based analytics and updates on AI-driven services and processes (Motamarri et al., Reference Motamarri, Akter and Yanamandram2020).

Doornenbal et al. (Reference Doornenbal, Spisak and Van der Laken2022) argue that AI can extend the leaders’ knowledge, particularly in complex circumstances. Knowledge and skills help leaders to apply AI-driven processes and assess the outcome (Akter et al., Reference Akter, Wamba, Mariani and Hani2021b). Such skills and knowledge equip leaders to serve customers using processes in an AI-driven environment (Motamarri et al., Reference Motamarri, Akter and Yanamandram2020). Training in a cutting-edge skill set enables leaders to handle varying service situations in AI-driven organizations, a key dimension of leaders’ AI-driven capability (Akter et al., Reference Akter, Wamba, Mariani and Hani2021b). Moreover, the evolution of AI technologies creates more efficient and effective ways to train and develop organizational leaders (Banks et al., Reference Banks, Dionne, Mast and Sayama2022). Motamarri et al. (Reference Motamarri, Akter and Yanamandram2020) find that regular exposure to training enhances leaders’ analytical capacity to serve better in the AI environment. They also argue that leaders are informed effectively regarding the updated AI-based skill set by participating in those trainings.

Adaptive capability

Leaders’ adaptive capability refers to the leaders’ ability to grasp and articulate the consequences of actions. Such ability enables leaders to assess the situation and ultimately choose paths to reach possible targeted outcomes (Campbell, Reference Campbell2021). Leaders with adaptive capability can quickly respond to evolving AI technologies, shifting business landscape, and adjust the organizational workforce dynamics accordingly to remain competitive. For example, adaptive capability enables leaders to mitigate the employees’ concerns regarding job security during AI adoption by assessing the requirements and reskilling employees for AI-augmented roles. The capacity to apply AI insights in decision-making, integrating human teams with AI effectively, inclination to experiment with AI-based innovations, and ability to align AI with regulatory and organizational standards indicate leaders’ adaptive capability in the AI environment. Adaptive capabilities assist leaders in creating and executing effective production, marketing, and human resources policies in the organization (Campbell, Reference Campbell2021). In an AI-driven context, leaders’ two key adaptive capabilities, problem-solving and decision-making are important. Problem-solving as the leaders’ adaptive capability is a core concern for organizational scientists. However, the digital era has vastly outpaced the traditional mechanisms to create organizational intelligence. Recent developments in technology, research, and organizational practice focus on leaders’ capability to use AL to create intelligent organizations (Kolbjørnsrud, Reference Kolbjørnsrud2024). Akter et al. (Reference Akter, Wamba, Mariani and Hani2021b) find problem-solving capacity a significant dimension of leaders’ adaptive capability. They argue that leaders make decisions and solve problems using analytics insights from a service analytics perspective. The problem-solving capability entails the leaders’ capacity to fix any problem when it occurs, the notion of using data over experience, and the scope for leaders to solve problems creatively (Motamarri et al., Reference Motamarri, Akter and Yanamandram2020). Intelligent digital and human actors collectively solve problems to create an intelligent organization (Kolbjørnsrud, Reference Kolbjørnsrud2024).

Decision-making is shown in literature as a noteworthy dimension of adaptive capabilities (Kim, Shin & Kwon, Reference Kim, Shin and Kwon2012). Now, businesses need to make more decisions in all areas of operations than ever before. A study by David Deming from Harvard Kennedy School shows that in 1960, 6% of total jobs required core decision-making in areas like diagnosing, prioritizing, and strategizing, reaching 34% by 2018. AI has successfully enabled organizations to increase the volume of decision-making. Besides increasing efficiency, AI also seems effective in decision-making by improving accuracy (Agrawal et al., Reference Agrawal, Gans and Goldfarb2022). However, increasing volume and accuracy are not enough because decision-making in one area of the organization impacts other areas of the organization. Organizational leaders must re-visit the whole value chain paradigm while designing any AI-based decision-making system (Agrawal et al., Reference Agrawal, Gans and Goldfarb2022). Generally, in AI environments, Raisch and Fomina (Reference Raisch and Fomina2022) argue that a probabilistic rather than deterministic decision-making method is more productive in the combined human-AI decision-making process. Among the three archetypes of leaders’ decision-making capability (Skeptic, Interactor, and Delegator), Meissner and Keding (Reference Meissner and Keding2021) find that being an interactor is the most effective way of deciding and solving problems in a collaborative human–machine landscape. Wamba et al. (Reference Wamba, Gunasekaran, Akter, Ren, Dubey and Childe2017) highlighted the selected dimensions of big data analytics capability, which can be aligned with the leaders’ decision-making capabilities in the AI environment. Such decision-making capability refers to the leaders’ attitude toward the impact of AI on making quicker decisions, jobs and productivity, cost, and timeliness of AI-based decisions (Wamba et al., Reference Wamba, Gunasekaran, Akter, Ren, Dubey and Childe2017).

Transformational capability

Leaders’ transformational capability is widely used in politics, sociology, and organizational behavior. Such capability refers to the leaders’ ability to transform employee performances beyond expectations (Khan, Rehmat, Butt, Farooqi & Asim, Reference Khan, Rehmat, Butt, Farooqi and Asim2020). In an AI-driven context, leaders’ transformational capabilities (Davenport & Bean, Reference Davenport and Bean2021; Davenport & Foutty, Reference Davenport and Foutty2021; Watson et al., Reference Watson, Desouza, Ribiere and Lindič2021) highlight the capacity of leaders to bring organization-wide AI-driven changes. Leaders with transformational capability bring fundamental changes by aligning AI with operations, business strategies and culture to foster innovation and maximize AI’s potential for the organization’s sustained competitive advantages. For example, the chief executive officer of a retail company can revolutionize the supply chain efficiency and customer experience by introducing AI-driven demand forecasting for inventory management and AI-powered tailored recommendations for personalized customer experiences. The capacity to clearly communicate AI-driven vision with organizational employees, align AI strategies with organizational goals, effectively navigate AI-driven changes within the organization, and manage uncertainties during the changes are the indicators of leaders’ transformational capability in the AI environment. Davenport and Mittal (Reference Davenport and Mittal2023) emphasize systematic AI applications across the key functions of an organization to enhance new business processes and data-driven decision–making. Likewise, AI should promote new business models, products, and service offerings. Alternatively, AI technology should ultimately transform all aspects of business.

Lugtu (Reference Lugtu2020) finds sensemaking, model innovation, and integrated execution to be leaders’ key AI-driven transformational capabilities. Sensemaking (Helfat & Peteraf, Reference Helfat and Peteraf2015) can be conceptualized in an AI-driven context as the leaders’ capability to capture AI initiatives’ distant and near future (Rafferty & Griffin, Reference Rafferty and Griffin2004). Model innovation refers to the leaders’ capability to rethink organizational issues novelly (Lugtu, Reference Lugtu2020). Such capability can be measured by how leaders think about old organizational issues, raise questions about existing practices, and challenge the basic assumption of such practices (Rafferty & Griffin, Reference Rafferty and Griffin2004).

Integration and reconfiguration capabilities, which are dynamic in nature, can also be regarded as leaders’ transformational capabilities (Helfat & Peteraf, Reference Helfat and Peteraf2015; Teece, Pisano & Shuen, Reference Teece, Pisano and Shuen1997). Wamba et al. (Reference Wamba, Gunasekaran, Akter, Ren, Dubey and Childe2017) conceptualize such capability as how leaders coordinate and control AI-driven organizational initiatives. Coordinating capability includes how leaders share AI-related knowledge and how AI issues are settled in functional and cross-functional meetings within the organization. Controlling capability includes leaders’ clear idea about AI responsibility within the organization, AI performance criteria, AI-driven business processes, and methods. Such capability also includes the leaders’ attitude toward appraising AI-based proposals, the cost-effectiveness of AI applications, and using detailed information regarding any AI applications, processes, and methods (Wamba et al., Reference Wamba, Gunasekaran, Akter, Ren, Dubey and Childe2017). Teece et al. (Reference Teece, Peteraf and Leih2016) bring leaders’ uncertainty-dealing capability as a key transformational leadership dimension. Such capability, in an AI-driven context, can be studied to the extent leaders presume the unpredictability of AI-driven changes will bring to the different units at the workplace, the consequential severity of such AI-driven changes, and having an idea of addressing those change consequences in AI-driven ways (Matsunaga, Reference Matsunaga2022).

Ethical resonance

Ethical resonance refers to the consistencies of human behavior that reflect core values in changing circumstances. Watson et al. (Reference Watson, Desouza, Ribiere and Lindič2021) focus on ethical concerns in AI-driven environments, which result in responsible leadership. Leaders with AI capability can better address the ethical issues related to AI adoption. A Deloitte study (2024) reveals that 89% of the top-level executives recognize the significance of robust governance structure and ethical framework required to implement AI in the organization. The same study also shows that 76% of the organizations focus on developing AI capability among the employees, while 63% of the organizations have extended their training initiatives up to the board of directors . Recently, AI experts at the TIME100 Impact Dinner in London (TIME Magazine, 2024) deliberated on the critical role of leaders’ AI capability in navigating ethical complexities related to AI integration. These examples indicate that leaders’ deeper understanding of AI enables them to implement better ethical practices.

From a leadership perspective in AI-driven organizations, moral judgment, trustworthiness, and systemic view emerge as key ethical dimensions. First, moral judgment means evaluating individual decisions by the standard of absolute truths. Business ethics evaluate the actions of leaders and organizations based on a moral perspective (Ferrell, Fraedrich & Ferrell, Reference Ferrell, Fraedrich and Ferrell2011). Bazerman and Gino (Reference Bazerman and Gino2012) focus on the influence of moral development on the ‘right or wrong’ perception of individual decisions. Moral judgment means how a leader resonates with the ethicality of individual decisions based on moral foundations, which appears to be a key outcome of responsible leaders in recent literature (Telkamp & Anderson, Reference Telkamp and Anderson2022). Zhang, Chen and Xu (Reference Zhang, Chen and Xu2022) found that AI decides with high competence but relatively low warmth, whereas humans decide with high warmth but relatively low competence. They conclude that AI is more capable of deciding on the utilitarian approach, and humans show higher capability in deciding on deontological approaches. Meissner and Keding (Reference Meissner and Keding2021) explore the role of leaders in AI discrimination and claim that moral dilemmas of AI-driven decisions should be the avenues of future research. Bruder (Reference Bruder2020) focuses on the autonomy of individuals for moral decision-making. However, Toth et al. (Reference Tóth, Caruana, Gruber and Loebbecke2022) view morality from a consequential perspective of individual decisions. They argue that the distribution and magnitude of consequences determine the morality of leaders’ decisions.

Second, leaders’ trustworthiness is critical for organizational success. Lacking trust hinders positive and productive collaboration in the organization (Lewis, Reference Lewis2021). Rebaie (Reference Rebaie2020) finds a relationship between trustworthy AI and trustworthy leadership. Leaders use AI insights to foster team effort and social learning processes to enhance the curiosity of team members. Moreover, leaders need to drive the cultural evolution in AI environments instead of letting technology drive the evolution. Otherwise, the risk of social resistance in uncertain and disruptive times would jeopardize organizational achievements (Rebaie, Reference Rebaie2020). In the AI context, leaders’ trustworthiness emerges as a significant ethical dimension that emphasizes a quick response to the organizational AI-related issues raised by the employees at the workplace (Lämsä & Keränen, Reference Lämsä and Keränen2020).

Third, leaders’ systemic view helps them to see the organization from the users’ perspective. Managers tend to view the organization from their hierarchical positions, whereas leaders intend to collaborate across the functional silos (Dunne, Eriksson & Kietzmann, Reference Dunne, Eriksson and Kietzmann2022). In an AI-driven context, a systemic view refers to the leaders’ capacity to view AI from a social system perspective by carefully evaluating the interests of multiple stakeholders involved (Crawford & Calo, Reference Crawford and Calo2016). Ghosh, Wilson, Burden and Dougherty (Reference Ghosh, Wilson, Burden and Dougherty2019) argue that organizations in AI environments will require human talent with systemic views of bridging development tools, infrastructure, programming languages, AI, and ML. Leaders must combine smart machines with human talents to ensure a new form of hybrid IT roles. Moreover, leaders must adapt their employees with relentless AI advancements based on ongoing learning from organizational transformations. Ghosh et al. (Reference Ghosh, Wilson, Burden and Dougherty2019) conclude that tomorrow’s successful leaders will design systems that can adjust to people rather than people to adjust the systems. A systemic view ensures stakeholders’ involvement and contemplates their consequences in all stages, including conception, design, implementation, and regulation (Crawford & Calo, Reference Crawford and Calo2016). Based on these insights, we propose that:

Responsible leadership

The concept of responsible leadership links the actions of senior executives and business performance with social responsibility and corporate sustainability based on ethical principles. Responsible leadership is a relational influence process between stakeholders and leaders to establish accountability in the organizational value-creation process (Maak, Pless & Voegtlin, Reference Maak, Pless and Voegtlin2016). Rooted in business ethics, responsible leadership is a moral leadership construct that assumes an extended stakeholder environment where the competing interests of both internal and external stakeholders are addressed in the long run. Ethics increasingly guides leaders in adopting the emerging responsible leadership approach. Ethical resonance is crucial to developing AI strategies in the organization that will ensure the societal greater good, upholding societal values and ethical standards. For example, the Los Angeles City Council is exploring ML to provide housing allocations for unhoused families by identifying systematic biases and flaws in the current system (Stern, Reference Stern2024). The city is using AI to engage community stakeholders and correct racial biases by demonstrating how ethical AI promotes social services fairness. Responsible leaders take an integrative approach to embed their actions with ethical reflection. They adopt a competitive approach with care for stakeholders. This integrative thinking highlights a systemic understanding of business in a society where the organization is part of a larger context (Pless & Maak, Reference Pless and Maak2005). Responsible leadership focuses on socially responsible and sustainable organizational transformation to generate positive social change (Pless & Maak, Reference Pless and Maak2004). Responsible leaders in an AI context will be challenged increasingly to resolve AI-driven uncertainties. Technological advancements from AI to robotics bring rapid transformations to businesses, which create new challenges for responsible leadership (Fernando & Bandara, Reference Fernando and Bandara2021). A study in 2020 revealed that Instagram’s algorithms prioritize users showing more skin in their photos. Consequently, the rest of the users remain out of Instagram’s organic reach. Clearview, a US firm, operates in 26 countries with police forces, law enforcement agencies, and governments by granting access to its database of three billion pictures. These pictures were taken from social media without consent. Even if they are not aware, business leaders are held accountable for the unethical outcomes in their organizations, and stakeholders want more socially responsible organizations (Samimi, Cortes, Anderson & Herrmann, Reference Samimi, Cortes, Anderson and Herrmann2020).

To address these issues in the AI environment, responsible leaders need ethical resonance that requires moral judgment, systemic view, and trustworthiness. Over the last few years, AI capabilities have gradually shifted from task accomplishment to relationship support (Tsai et al., Reference Tsai, Marshall, Choudhury, Serban, Hou, Jung and Yammarino2022). As an integral dimension of relational intelligence, ethical resonance focuses on building trustful relationships with diverse stakeholders (Maak & Pless, Reference Maak and Pless2006). While the general leadership view is focused on internal stakeholders, financial performance, and accountability toward shareholders, responsible leadership correlates the actions with the interests of multiple stakeholders based on ethical principles for sustainable business success (Maak & Pless, Reference Maak and Pless2006). Based on this discussion, we propose:

Kompella (Reference Kompella2022) highlights responsible AI as part of an organization’s risk management practices that help minimize situations that harm organizational reputation. The responsible use of AI also facilitates better customer service and creates the same platforms for all vendors within the AI procurement system (Kompella, Reference Kompella2022). Such concerns pertinently raise the acute need for leaders of AI organizations to lead responsibly. International Telecommunication Union calls for a responsible AI framework to strengthen confidence in deploying AI-based solutions and trusting outputs. The framework highlights the importance of appropriate leadership to ensure effective AI-based initiatives and to monitor AI-based outcomes (ITU, 2017). Responsible leadership facilitates generating responsible organizational outcomes. Responsible leaders not only adapt to newer technological improvements but also anticipate possible problems with such new adaptations and find ways to resolve those problems (McAfee, Goldbloom, Brynjolfsson & Howard, Reference McAfee, Goldbloom, Brynjolfsson and Howard2014).

Leaders’ AI-driven capability influences navigating responsible leadership in AI-driven organizational environment. AI can promote more inclusive and equitable decision-making. Leaders with greater AI competencies can integrate AI throughout the organization more strategically and responsibly. For example, leaders with AI-driven capability can adopt AI to mitigate biasedness in the organizational recruitment system. Designing AI platforms that can anonymize names and genders helps to prevent unconscious biases in hiring, and thus promotes a more equitable and diverse workforce within the organization (van Esch, Cui & Heilgenberg, Reference van Esch, Cui and Heilgenberg2024). Several scholars have studied human–machine interactions in terms of responsible leadership. Agrawal et al. (Reference Agrawal, Gans and Goldfarb2017) focus on a co-creation perspective by adopting AI-system-based prediction and note that human-capability-based judgment will lead organizations toward the right strategic directions. Garfinkel, Matthews, Shapiro and Smith (Reference Garfinkel, Matthews, Shapiro and Smith2017) mention that leaders must take responsibility for monitoring AI-based operations. Although AI largely predicts tasks in uncertain circumstances, prediction itself is not the decision (Agrawal et al., Reference Agrawal, Gans and Goldfarb2017). Wilson and Daugherty (Reference Wilson and Daugherty2018) mention that leaders must acquire fusion skills to work in a collaborative human–machine landscape (Lichtenthaler, Reference Lichtenthaler2018). Due to increased AI applications, Kolbjørnsrud et al. (Reference Kolbjørnsrud, Amico and Thomas2017) find leadership roles more creative, judgmental, flexible, and tacit knowledge oriented. Davenport (Reference Davenport2016) concludes that a collaboration of human–machine cognitive intelligence will ultimately be translated into an integrated and responsible strategic approach for competitive organizations. Such a collaborative leadership role is reflected as a trainer, explainer, and sustainer (Wilson, Daugherty & Bianzino, Reference Wilson, Daugherty and Bianzino2017). Moreover, Davenport and Kirby (Reference Davenport and Kirby2016) suggest that based on the facts and contexts, the capacity of leaders for sensemaking in the big picture will result in responsible AI initiatives and sustainable organizational performances. Thus, we propose: