Abstract: Artificial Intelligence (AI) is increasingly capable of performing leadership-related functions—analysing data, allocating tasks, and even coaching team members—alongside human managers. Yet organisations lack a clear framework for augmented leadership in which humans and AI systems collaborate as joint leaders of teams. This paper proposes a novel conceptual framework for leadership in human–AI hybrid teams, outlining key roles that AI can play (as analytical partner, operational co-pilot, skills coach, and compliance guardian) and how human leaders can strategically integrate these AI contributions. We synthesise literature on algorithmic management, human–AI interaction, and leadership theory to identify critical factors (e.g. trust, shared mental models, ethical accountability) that determine the success of AI augmentation. The framework yields a set of propositions about the balance of responsibilities between human and AI co-leaders and the conditions under which hybrid teams outperform purely human-led teams. In discussion, we contrast our approach with traditional leadership models and highlight how it extends shared leadership and sociotechnical systems theory into the AI context. We then derive practical implications for managers—such as strategies for building trust in AI and retaining human oversight—and outline directions for future research to empirically evaluate and refine the proposed model. By treating AI not as a threat or a mere tool but as a collaborative co-leader, this work aims to advance scholarly understanding and guide organisations in leveraging AI to elevate team performance and innovation.
Key Words: artificial intelligence; leadership; human–AI collaboration; hybrid teams; trust
Introduction:
Artificial Intelligence (AI) is transforming workplaces by taking on tasks traditionally performed by human managers. Globally, 75% of executives report that AI will substantially change their leadership processes by 2025 (World Economic Forum, 2023), and India has emerged as a leader in AI adoption – 92% of Indian companies use AI in some capacity, the highest rate worldwide (KPMG, 2025; Chougule, 2025). Early corporate experiments illustrate this trend. In 2023, IBM’s CEO announced that an AI “AskHR” agent had automated 94% of routine HR queries, effectively eliminating several hundred human HR roles, while those staff were redeployed into new “critical thinking” positions (Kapoor, 2023). This deployment was credited with significant productivity gains (IBM’s COO attributed a $3.5 billion improvement to AI augmentation) (Kapoor, 2023). Likewise, Salesforce has begun using AI “gig agents” to handle surges in customer service, with CEO Marc Benioff suggesting that AI can temporarily take over certain team tasks during peak periods (Eaton, 2025). These cases show that AI systems are no longer mere tools but are starting to act as quasi-colleagues or team members – performing coordination, analysis, and decision support alongside humans. Early adopters report significant efficiency gains; for example, IBM observed faster query resolution and improved HR productivity (Kapoor, 2023). On the other hand, integrating AI into teams has raised challenges around trust and role clarity, as employees in some cases feared its decisions or felt uncertain about how to collaborate with a non-human “teammate” (Rosenblat & Stark, 2016; Möhlmann & Zalmanson, 2017). This highlights the need for new leadership approaches to ensure AI’s benefits are realised without undermining team morale or clarity.
Existing research provides insights on this emerging phenomenon. Studies of algorithmic management show that AI-driven systems can indeed take on managerial functions like task allocation and performance evaluation, yielding efficiency gains but sometimes harming worker morale and autonomy (Rosenblat & Stark, 2016; Möhlmann & Zalmanson, 2017). In contrast, research on AI augmentation emphasises using AI to support – not replace – human decision-making. Wilson and Daugherty (2018) introduced the concept of “collaborative intelligence,” arguing that AI and humans have complementary strengths and achieve the best results when working in tandem. Empirical evidence supports this: executives who used AI decision support combined with their own judgement outperformed those relying on intuition alone and those following AI blindly (Bazerman, 2020). Similarly, doctors assisted by AI diagnostic tools made more accurate decisions than unaided doctors, yet human oversight remained crucial to catch nuance and prevent mistakes (Raisch & Krakowski, 2021). These findings align with what Raisch and Krakowski (2021) call the automation–augmentation paradox: organisations must pursue the efficiency of AI automation and the adaptability of human judgement simultaneously, since leaning too far toward either extreme creates risks – over-automation can reduce flexibility and trust, while over-reliance on humans can forgo AI’s speed and consistency (Raisch & Krakowski, 2021). Therefore, scholars increasingly advocate a “both/and” approach that deliberately combines AI and human inputs in management (Seeber et al., 2020; Jarrahi, 2018).
Research Gap:
Despite these early insights, current literature lacks a cohesive framework guiding how humans and AI can jointly lead teams. Prior studies have examined isolated issues – for example, building trust in AI teammates (Glikson & Woolley, 2020), AI’s influence on specific leadership behaviors (Crawford, Laird & Moss, 2023), or employee responses to algorithmic decisions (Rosenblat & Stark, 2016) – but these have not been integrated into a unified model of human–AI shared leadership. Traditional leadership theories assumed all team members were human and provide little guidance for scenarios where a non-human intelligent agent participates in leadership (Rosenberg, Tschopp & Moerschell, 2022). As a result, managers experimenting with AI in leadership roles are often operating on guesswork or vendor advice, rather than a solid evidence-based framework. Fundamental s remain: Which leadership tasks can be delegated to an AI, and which must stay human-led? How should human leaders and team members adapt their practices when a machine becomes part of the team? Under what conditions does human–AI co-leadership improve team performance or, conversely, create new issues?
Purpose and Contribution:
The aim of this paper is to address these gaps by developing a conceptual framework for augmented leadership in human–AI hybrid teams. We ground our framework in a broad review of literature (leadership theory, technology acceptance, trust in automation, and sociotechnical systems) and real-world observations, ensuring it rests on established concepts rather than conjecture. We explicitly delineate how our framework extends prior work. Essentially, it can be viewed as an evolution of shared leadership theory (Pearce & Conger, 2003) where the sharing occurs between a human and an AI system rather than exclusively among humans. This introduces novel dynamics (e.g., calibrating trust in an algorithmic teammate, defining clear human vs. AI authority) not addressed by earlier models. The framework defines key leadership domains and proposes specific role allocations for the AI and human leader in each, accompanied by conditions that enable the partnership to succeed. We then derive theoretical propositions and practical guidelines.
The contributions of this work are twofold. Theoretically, it provides an integrated model of human–AI co-leadership, bridging classical leadership theories and modern human–AI collaboration insights. It incorporates constructs from multiple perspectives (shared leadership, trust and technology acceptance models, adaptive leadership, and sociotechnical systems thinking) to articulate how leadership responsibilities can be distributed between humans and AI. This contributes to leadership theory by extending concepts like distributed/shared leadership and LMX (leader–member exchange) to include non-human agents, and to technology management theory by highlighting the social and cultural factors affecting AI’s integration. Practically, it offers a framework that organisations and managers can use as a playbook when implementing AI in team leadership. This includes recommendations on dividing tasks with AI, training human leaders and team members for augmented collaboration, maintaining team trust and clear communication with an AI partner, and ensuring ethics and accountability. These recommendations are illustrated with examples and underpinned by research, providing actionable insights for companies navigating the human–AI hybrid team frontier.
In the sections that follow, we first review the pertinent literature (Section 2) and identify specific gaps that motivate our work. We then outline our research approach (Section 3). Next, we present the Augmented Human–AI Co-Leadership Framework (Section 4), detailing its domains, roles, and assumptions and providing a conceptual diagram and summary table for clarity. Finally, we discuss implications for theory and practice and suggest directions for future research (Section 5).
2. Literature Review
The emergence of AI in team settings lies at the intersection of multiple research streams. We organise our review into three key themes: (2.1) AI’s evolving role in management from automation to augmentation; (2.2) human–AI team dynamics including trust, communication, and alignment; and (2.3) specific leadership challenges introduced by AI in teams (authority, ethics, development). This review situates our framework within existing knowledge and identifies open questions that it must address.
2.1 AI’s Role in Management: From Automation to Augmentation
Early organisational uses of AI emphasised automation, with algorithms executing structured tasks and managerial functions to reduce costs and improve consistency. On gig platforms like Uber, for example, algorithms assign rides and rate drivers, acting as managers by enforcing rules uniformly (Rosenblat & Stark, 2016). Such systems can increase efficiency and control, but they often undermine flexibility and human aspects of leadership, leaving workers feeling reduced to “cogs in a machine” (Möhlmann & Zalmanson, 2017). For instance, Uber drivers have reported frustration with the algorithm’s rigid decisions and lack of human understanding (Rosenblat & Stark, 2016). These experiences highlight the limits of pure automation in roles requiring judgement and empathy.
In contrast, many scholars and practitioners now advocate an augmentation approach – using AI to assist and enhance human managers rather than replace them (Davenport & Kirby, 2016; Jarrahi, 2018). This “collaborative intelligence” model posits that humans and AI have complementary strengths that can be combined (Wilson & Daugherty, 2018). Empirical findings underscore the benefits of this approach. Brynjolfsson et al. (2011) found that companies combining algorithmic analytics with human judgment achieved higher performance than those relying on just human intuition or just algorithms. In one study, executives who consulted AI-based forecasts but made final decisions themselves achieved better outcomes than those who ignored AI or followed it blindly (Bazerman, 2020). Similarly, in high-stakes contexts like healthcare, AI diagnostic tools plus human experts tend to outperform either alone (Raisch & Krakowski, 2021). These examples illustrate that maximum effectiveness often comes from balanced human–AI collaboration, not extreme automation.
Raisch and Krakowski (2021) formalise this idea as the automation–augmentation paradox, arguing that the greatest gains come from carefully combining automation with human adaptability. Lean entirely on automation and you may get consistency and speed at the cost of creativity, trust, and resilience; rely solely on humans and you miss opportunities for data-driven improvements and scalability. The literature thus recommends a balanced distribution of roles. Seeber et al. (2020) discuss treating “machines as teammates” with defined responsibilities in teams, rather than as mere tools or replacements. Where early discussions focused on whether AI would replace managers, the focus is now on how AI and managers can collaborate effectively – a fundamental shift requiring new frameworks.
Despite these insights, the practical how-to of sharing leadership with AI remains unclear. Each organisation or team tends to experiment in isolation, as no comprehensive model currently guides them in choosing which tasks to augment or how to maintain team morale and performance through the transition.
2.2 Human–AI Team Dynamics: Trust, Communication, and Shared Mental Models
Researchers consistently find that trust is pivotal in human–AI teams (Hoff & Bashir, 2015; Glikson & Woolley, 2020). Trust in an AI system refers to team members’ willingness to rely on the AI’s output in uncertain situations (Madsen & Gregor, 2000). Too little trust means ignoring useful AI inputs; too much trust means uncritically following the AI into errors. “Calibrated” trust – trusting an AI to the extent it is reliable – is the goal (Lee & See, 2004).
How can calibrated trust be fostered? Studies show transparency and explainability are key (Glikson & Woolley, 2020). For instance, Lyons et al. (2022) found that when an AI support system explained its recommendations, human operators followed its advice appropriately – trusting it in routine cases but applying skepticism in anomalies. Without explanations, operators either under-trusted (ignoring good advice) or over-trusted (following flawed advice blindly). Thus, giving users insight into why the AI suggests something helps them trust it in the right measure. Over time, consistent performance and clear communication about the AI’s limits build a shared mental model of the AI’s role (Lee & See, 2004).
Besides individual trust, shared understanding at the team level is vital. In human-only teams, members perform better when they share mental models of goals, tasks, and each other’s roles (Mathieu et al., 2000). In a hybrid team, the mental model must include the AI agent’s role, capabilities, and boundaries (Seeber et al., 2020). Team members need clarity on what the AI will do and not do; the AI must be set up (by its designers or the team) to reflect relevant team norms and context. Misalignment leads to confusion or conflicts – e.g., if an AI doesn’t “know” about an informal exception the team uses, it might flag normal behavior as a violation, annoying the team. Kamar et al. (2019) found that teams explicitly briefed on an AI co-ordinator’s role and rules outperformed those left to figure it out themselves. Such explicit role definition helps team members know when to rely on the AI vs. when to take charge, thereby smoothing collaboration.
Effective communication is also central. AI systems usually communicate via dashboards, alerts, or suggestions, lacking the nuance of human conversation. If an AI offers recommendations without context, it may be perceived as a black box and breed distrust or misinterpretation (Nourani et al., 2020). Therefore, human team members and leaders must adjust their communication strategies. A leader might discuss the AI’s suggestion openly: “Our project assistant flagged a risk of delay; I think it’s because it noticed our testing backlog doubling”. By verbalising the AI’s input and integrating it into team dialogue, the leader not only validates the AI’s contribution but also clarifies it for everyone (Glikson & Woolley, 2020). Meanwhile, team members should be encouraged to voice uncertainties or disagreements with AI outputs.
This ties into psychological safety – the degree to which team members feel safe taking interpersonal risks (Edmondson, 1999). In a hybrid team, psychological safety must extend to interactions involving the AI. Team members should feel comfortable saying, “I think the AI’s suggestion might not apply here because [contextual reason],” without fear of ridicule or reprisal. Leaders can foster this by explicitly encouraging scrutiny of the AI’s decisions and reminding the team that challenging the AI is acceptable and even necessary (Rosenberg et al., 2022). If employees believe they must silently defer to the AI or their leader’s endorsement of it, they may suppress valid concerns, leading to undetected errors or resentment. Conversely, if they feel safe providing feedback on the AI’s outputs, they become active collaborators in improving the human–AI team’s performance. In summary, calibrated trust, role clarity, and open communication have emerged as foundational ingredients for successful human–AI teaming. Our framework will incorporate these as prerequisites for effective co-leadership.
2.3 Leadership Challenges in Human–AI Hybrid Teams
Integrating AI into team leadership brings specific challenges beyond those of typical teams. We highlight three: (a) balancing authority and autonomy, (b) ensuring ethics and accountability, and (c) preserving team members’ motivation and development.
(a) Balancing Authority and Autonomy: If too much decision-making power is given to an AI too quickly, employees can become uncomfortable or resistant – feeling they have lost agency or fearing the AI might make unfair choices (Glikson & Woolley, 2020). Conversely, if an AI tool is under-utilised (needing human sign-off on every action), its speed and consistency benefits are squandered (Raisch & Krakowski, 2021). Leaders therefore face the task of defining clear decision boundaries: e.g., an AI might autonomously assign routine tasks or handle schedule adjustments within a one-day range, but anything beyond that requires human approval. This approach corresponds to a “human-on-the-loop” model (Sheridan, 2019) – the AI can act on its own in low-stakes, well-defined scenarios, while the human supervisor remains available to intervene in complex or novel situations. Such conditional delegation has been effective elsewhere: for example, autopilot systems fly aircraft under normal conditions, but pilots take control during takeoff, landing, or turbulence, combining efficiency with human judgment (Endsley, 2017). In teams, a human leader might similarly let an AI reorder tasks for optimal workflow, but step in to manually manage a major crisis or exception. The key is explicitness: team members and the AI itself (via programming) must know who has authority over what. This clarity preempts confusion and frustration, and helps team members trust the AI within its domain.
(b) Ensuring Ethics and Accountability: Introducing AI into leadership tasks complicates questions of responsibility. The fundamental principle must be that human leaders remain accountable for outcomes – no matter how autonomous the AI, the organisation cannot abdicate responsibility to a machine (OECD, 2019). In practice, this means the human leader must actively oversee the AI’s important decisions, put checks in place, and address potential issues proactively. For instance, if an AI-driven performance review system recommends firing certain employees, the leader should carefully review these cases (with their context) rather than simply rubber-stamping them. Leaders must also watch for bias: AI systems can inadvertently perpetuate historical biases present in their training data (Buolamwini & Gebru, 2018). If an AI’s suggestion or action raises ethical questions – e.g., it appears to systematically disadvantage a group of employees – the leader has an obligation to investigate and correct that pattern (Zhang, 2021; Martin, 2019). In our framework, the human leader acts as ethical gatekeeper: the AI may monitor for issues (as we discuss later), but the human must interpret any flags and ensure decisions align with organisational values and fairness standards. This hands-on approach preserves morality and trust: employees see that even though an algorithm is involved, their leader will not blindly follow it into ethically questionable territory. Accountability, therefore, is maintained and even enhanced – the leader now has both their own decisions and the AI’s decisions in their purview (Raisch & Krakowski, 2021).
(c) Preserving Team Motivation and Development: A potential unintended consequence of AI integration is that it might diminish employees’ opportunities for growth or satisfaction if not managed well. There is concern about de-skilling: reliance on AI for complex tasks might lead humans to lose practice and, over time, lose expertise (Susskind & Susskind, 2022). For example, if an AI scheduling assistant always resolves conflicts, junior team members might not learn how to handle scheduling challenges themselves. Leaders should mitigate this by ensuring that the team continues to engage in critical thinking and skill development. One strategy is to use AI as a learning tool – e.g., encouraging team members to examine the AI’s analyses and solutions to learn new techniques or insights. Another key aspect is maintaining and even boosting job satisfaction and engagement. Huang and Rust (2021) found that when AI takes over tedious tasks, it can increase employee satisfaction by allowing them to focus on more creative and meaningful work (provided they still feel valued in the process). Leaders should highlight this benefit: e.g., “By automating reporting, we can spend more time brainstorming new ideas.” They also need to communicate a positive narrative: AI is here to augment your capabilities and remove drudgery, not to replace you. This can alleviate anxiety and help team members see AI as a partner. Change management best practices emphasize the importance of such framing (Kotter, 1996). The leader’s own behaviour matters too – by actively helping team members adapt and providing reassurance (for example, praising team members’ unique contributions that the AI cannot make), they can ensure that the introduction of AI does not erode morale or cause disengagement.
In summary, human–AI hybrid teams present unique leadership challenges: delegating authority carefully to the AI, maintaining ethical leadership and accountability, and sustaining human team members’ development and well-being. These challenges motivate our framework’s focus on structured role allocation, oversight, and culture.
3. Research Methodology
Scope and Assumptions:
We focus on knowledge-intensive, innovative team contexts (e.g., technology product development, R&D, or IT project teams) where AI tools are being implemented to support leadership tasks. These contexts are early adopters of AI for team management, and they involve complex, non-routine work – conditions under which augmented leadership can be particularly impactful (Jarrahi, 2018). We limit our consideration to currently feasible narrow AI systems (project management assistants, analytics and decision support tools, chatbots for coaching or communication, etc.), rather than speculative artificial general intelligence or fully autonomous “robot managers.” We assume human leaders remain ultimately responsible and accountable for team outcomes and that AI is introduced to complement human skills, not to compete with or completely replace human leaders. We also assume teams that have baseline tech literacy and openness to adopting new digital tools, since those are likely needed for effective human–AI collaboration.
Methodological Approach:
This is a conceptual, theory-building study, not an empirical analysis. We performed an integrative literature review, following guidelines by Torraco (2005), to gather and synthesise relevant findings from academic research and industry reports. This process allowed us to identify critical variables (like trust, transparency, etc.) and theoretical lenses (e.g., shared leadership, sociotechnical theory) that inform our model. We also used a case-informed approach (Siggelkow, 2007), referencing real examples (from IBM, Salesforce, etc.) as “proof-of-concept” illustrations of how aspects of our framework manifest in practice. The research is inductive: we bring together findings from multiple disciplines to propose a unified framework, then articulate propositions that can be tested in future research. The methodology is akin to developing a new model through conceptual integration. Our approach was to iteratively refine the framework by checking it against known evidence and logical consistency.
We do not present new primary data here, but our aim is to provide clear, testable claims (in the form of propositions) that can be examined in future empirical studies. For example, future researchers could survey teams with and without AI co-leadership to test whether our proposed mediators (like team trust in AI) indeed influence performance, or run experiments in simulated team tasks to compare traditional vs. augmented leadership structures. In Section 5, we suggest such avenues, demonstrating how the conceptual constructs identified (trust in AI, leader adaptability, etc.) might be operationalised.
4. Augmented Human–AI Co-Leadership Framework
Building on the above, we propose an integrative framework for human–AI co-leadership in teams. The framework spans four key leadership domains: (1) Analytical Decision-Making, (2) Task and Process Management, (3) Team Development and Support, and (4) Monitoring and Governance. These domains correspond to classic leadership functions (e.g., planning/deciding, coordinating, supporting, controlling) seen in various taxonomies (Yukl, 2012; Bass, 1985). Within each domain, we specify an AI co-leader role that taps into typical AI strengths, and a complementary human leader role focusing on areas requiring human qualities. This division is designed to optimise overall team leadership by having each partner (human or AI) focus on what it does best. Surrounding all four domains are the crucial enabling conditions identified in Section 2 (calibrated trust, role clarity, transparency, etc.), which we treat as prerequisites for success.
Figure 1 provides a conceptual diagram of the framework, and Table 1 summarises the domains and roles at a glance. The subsequent text describes each domain and role division in detail.
Domain 1 – Analytical Decision-Making.
AI as Cognitive Partner; Human as Strategist & Final Arbiter. This domain involves strategic analysis and evidence-based decision-making. Here, the AI serves as a Cognitive Partner, providing the human leader and team with high-speed data analysis, pattern recognition, and unbiased recommendations. Modern AI can rapidly analyse large datasets, identify trends or anomalies, run simulations, and generate options, which significantly augments the team’s intelligence gathering and problem-solving ability (Jarrahi, 2018). The human leader’s complementary role is the Strategist & Final Decision-Maker – they integrate the AI’s insights with tacit knowledge, experience, and understanding of broader context to make final decisions. The human has the last word, especially on ambiguous or value-sensitive issues where human judgement, creativity, and intuition are crucial (Ajzen, 1991). For example, an AI might analyse market data and recommend a strategic pivot; the human leader considers this analysis alongside factors like organisational values and risk tolerance, then decides whether to act on it. Enabling Conditions: Transparency and trust are vital in this domain. The AI should provide interpretable outputs (e.g., data visualisations or explanations) so the leader and team trust its recommendations (Lyons et al., 2022). The leader, in turn, should openly discuss how the AI’s insights inform decisions, reinforcing that while the AI is a valuable adviser, the human is accountable for the final call. This division of labour combines AI’s analytic prowess with human holistic judgement, reflecting the principle that humans and AI together can make better decisions than either alone (Bazerman, 2020; Mintzberg, 1973).
Domain 2 – Task & Process Management.
AI as Operational Co-Pilot; Human as Orchestrator & Exception-Handler. This domain covers the execution and coordination of work: assigning tasks, scheduling, tracking progress, and managing workflow. The AI can function as an Operational Co-Pilot by automating routine coordination and monitoring tasks: for instance, assigning tasks to team members based on workload (akin to how Uber’s system dispatches drivers), sending deadline reminders, or flagging resource conflicts according to rules. AI excels at reliably handling such structured, repetitive processes without fatigue, ensuring consistency and freeing humans from micromanagement. The human leader, as Orchestrator & Exception-Handler, oversees the overall operation, monitors for issues beyond the AI’s scope, and intervenes to handle exceptions, complex trade-offs, or conflicts that require human flexibility and negotiation (Hackman & Walton, 1986). For example, an AI scheduling assistant might reallocate tasks to optimise workflow; the human leader monitors these changes and steps in if a client-facing deadline needs reprioritising due to relationship considerations the AI isn’t aware of. Enabling Conditions: Clear authority limits must be set so everyone understands what the AI can decide autonomously. The leader should define and communicate, for example, that the AI will handle internal task assignments, but any client scope changes must be approved by the human. The team should also trust the AI’s processes – built over time through seeing its reliability in handling mundane tasks (Glikson & Woolley, 2020). Meanwhile, the human leader should regularly review the AI’s decisions to ensure they remain aligned with broader objectives and to adjust the AI’s parameters if needed. When properly executed, this co-leadership in operations can dramatically increase efficiency and consistency (e.g., no waiting for a manager to manually assign every small task), while maintaining the human flexibility and judgement needed for novel situations (Sheridan, 2019; Endsley, 2017).
Domain 3 – Team Development & Support.
AI as Skills Coach; Human as Mentor & Culture Builder. This domain entails guiding and supporting team members’ growth and maintaining a healthy team climate. The AI can act as a Skills Coach or Tutor, offering personalised, on-demand training and feedback. With access to vast knowledge bases, AI chatbots or assistants can answer questions, provide how-to guidance, or critique a piece of work (e.g., an AI coding assistant suggesting improvements). AI can also give data-driven performance feedback – for example, analysing call centre agents’ conversations and offering tips (Bock, Holtgrewe & Kopp, 2022). However, current AI cannot replace human empathy or truly inspire people (Huang & Rust, 2021), so the human leader remains the Mentor & Culture Builder. The leader provides emotional support, career mentoring, motivates individuals, and sets the tone for an inclusive, positive team culture. In practice, the AI can help employees with day-to-day skill queries and objective feedback (Johnson et al., 2020), while the human leader does what only humans can: showing understanding, nurturing relationships, and aligning the team’s values and morale. Enabling Conditions: Trust and a supportive climate are crucial here. Team members should see the AI as a non-judgmental assistant, not an invasive monitor. That requires the leader to introduce the AI’s coaching role in a positive way (e.g., “This tool is here to help you improve at your own pace”) and ensure it is used for development rather than surveillance (Singh et al., 2021). The leader can reinforce this by, for example, using the AI’s feedback in coaching conversations and framing it as helpful tips. If an AI flags a skill gap or performance issue, the leader addresses it constructively and contextually – providing encouragement and understanding beyond what the AI can offer. Psychological safety is again important: employees should feel free to critique or correct the AI’s feedback if it seems wrong, without fear. By effectively combining AI’s knowledge with human mentorship (where AI handles factual tutoring and humans handle motivation and personal growth), teams can achieve faster skill development while maintaining high engagement (Huang & Rust, 2021).
Domain 4 – Monitoring & Governance.
AI as Ethical & Performance Sentinel; Human as Accountability Partner & Culture Guardian. The final domain involves monitoring team activities to ensure performance and ethical standards, and taking corrective action when needed. AI can function as an Ethical & Performance Sentinel – continuously scanning project metrics, communications, or decisions for any issues. For example, an AI might analyse code contributions to flag potential quality problems or deviations from best practices, or monitor decision patterns to alert if there’s potential bias or compliance risk (Buolamwini & Gebru, 2018). The human leader serves as the Accountability Partner & Culture Guardian. When the AI flags something (e.g., a risk, a rule violation, or a sign of team imbalance), the human leader examines it in context and decides on the response, bringing in nuance and values that AI lacks. Crucially, the leader upholds ethical standards and ensures any AI-driven insights are used fairly and responsibly. Enabling Conditions: Transparency and fairness are paramount in this domain. The leader should be upfront with the team about what the AI is monitoring and why (e.g., quality control, safety, fairness) to prevent a “big brother” atmosphere. And if the AI detects an issue, the leader must handle it in a way that upholds trust – for example, investigating impartially and discussing it with the team rather than imposing punitive measures solely on the AI’s output (Rosenberg et al., 2022). The leader’s active role keeps the AI’s monitoring function tethered to human judgement and organisational values (Brown et al., 2005; Martin, 2019). For instance, if an AI flags that certain voices are dominating team discussions, the leader might facilitate more inclusive meetings rather than simply silencing individuals per an algorithmic suggestion. This approach ensures that technology serves as a tool for improvement, while human leaders remain the arbiters and exemplars of the team’s culture and ethics. Maintaining human responsibility also aligns with emerging AI governance principles that stress human oversight and accountability for AI decisions (OECD, 2019).
Table 1. Human–AI Co-Leadership Roles Across Key Leadership Domains
| Leadership Domain | AI’s Role (Co-Leader) | Human Leader’s Role |
|---|---|---|
| Analytical Decision-Making – Strategic planning & problem-solving | Cognitive Partner: Processes large data sets, identifies patterns, and provides evidence-based recommendations to support team decisions. Strengths: Speed, unbiased analysis, breadth of information. | Strategist & Final Arbiter: Integrates AI’s data-driven insights with context, intuition, and creative judgement; makes final decisions and remains accountable for outcomes. Strengths: Big-picture vision, values-driven judgement, responsibility. |
| Task & Process Management – Execution & coordination | Operational Co-Pilot: Manages routine task assignments, scheduling, and progress tracking; optimises workflow and flags issues according to predefined rules. Strengths: Consistency, 24/7 vigilance, efficiency in structured tasks. | Orchestrator & Exception-Handler: Oversees AI-driven operations at a high level; intervenes when complex judgment, unforeseen changes, or conflicts require human flexibility; maintains overall alignment with objectives. Strengths: Adaptability, conflict resolution, holistic oversight. |
| Team Development & Support – Mentoring & team climate | Skills Coach & Tutor: Provides personalised, just-in-time training, information, and feedback to team members (e.g., answering questions, suggesting improvements). Strengths: Vast knowledge access, immediate availability, objective feedback. | Mentor & Culture Builder: Delivers emotional support, motivation, and career guidance; contextualises and humanises AI feedback; fosters a learning culture and psychological safety within the team. Strengths: Empathy, inspiration, morale-building, ethical guidance. |
| Monitoring & Governance – Standards & ethics | Ethical & Performance Sentinel: Continuously monitors team actions and outputs for potential risks, quality issues, or policy violations; flags anomalies for review. Strengths: Impartiality, constant vigilance, pattern detection across data. | Accountability Partner & Culture Guardian: Reviews AI flags with judgement and situational awareness; takes responsibility for final decisions; ensures fair, ethical responses and maintains team trust and values. Strengths: Moral reasoning, accountability, contextual decision-making. |
5. Discussion and Implications
The proposed Augmented Human–AI Co-Leadership Framework offers a structured approach to integrate AI into team leadership. We discuss its theoretical significance, practical implications, limitations, and suggest future research directions to test and refine the model.
5.1 Integration with Existing Theories:
This framework extends core concepts of leadership theory into the context of intelligent machines as team participants. It can be viewed as a new form of shared leadership (where leadership roles are distributed among multiple agents), now including a non-human agent (Pearce & Conger, 2003). It reflects elements of transformational leadership in Domain 3 (the human leader focuses on inspiration and individual consideration, tasks beyond AI’s reach) while leveraging AI for transactional elements like monitoring and routine coordination (Burns, 1978). It resonates with Leader–Member Exchange (LMX) theory in that the human leader must build a working relationship even with the AI – calibrating their interactions much as they would with a team member – to achieve effective “exchange” (Graen & Uhl-Bien, 1995). By integrating perspectives from technology acceptance models (e.g., highlighting trust, perceived usefulness, and facilitating conditions as in Venkatesh, Thong & Xu, 2012) and sociotechnical systems theory (Trist, 1981), our framework underscores that successful AI deployment is as much about social and organisational design as it is about technological capability. In short, it broadens leadership theory to account for hybrid agent teams, providing a vocabulary and set of constructs (like “AI cognitive partner” or “human accountability guardian”) for analysing leadership when one “member” of the leadership structure is artificially intelligent.
5.2 Addressing Key Challenges:
The framework directly addresses the challenges of co-leadership. Regarding authority balance, it provides a blueprint (via the domain-specific role allocations) for dividing decision rights and ensuring humans stay “on the loop” (Sheridan, 2019). For ethics and accountability, it makes clear that human leaders must remain answerable and places them firmly in the role of ethical arbiters (Martin, 2019; OECD, 2019). For human development and motivation, it preserves significant human-led roles (strategic decisions, mentoring, cultural leadership) that keep humans central and valuable, while using AI to offload the less engaging tasks – aligning with theories of job enrichment and motivation (Herzberg, 1968; Huang & Rust, 2021). This structured approach is intended to avoid common pitfalls such as unclear role boundaries (leading to confusion) or the erosion of team trust and skills. For instance, a potential problem is de-skilling: our framework counters that by recommending leaders to have team members actively engage with AI insights (thus learning from them) rather than just passively deferring to the AI.
5.3 Managerial Implications:
For practitioners, this framework serves as a guide to implement AI in team leadership roles responsibly and effectively. Key steps include:
- Role Definition & Clarity: Clearly delineate what the AI is responsible for and what remains with the human leader. Document these decisions and communicate them to the team (and adjust them as needed). For example: “Our AI scheduling assistant will handle routine task assignments and send reminders; I will handle any major changes or prioritization of tasks.” This prevents ambiguity and sets expectations.
- Leader Training in AI Oversight: Ensure human leaders are prepared to manage and collaborate with AI. This means training in understanding the AI’s functionalities and limitations – e.g., how it makes decisions, what data it uses – so they can effectively supervise it and integrate its outputs (Venkatesh et al., 2012). Also, leaders may need to develop new skills like data literacy or basic knowledge of AI systems to maintain credibility and competence in this new partnership.
- Building Team Trust in AI: Start small to build confidence. Perhaps deploy the AI in a pilot project or for limited functions initially, and share its successes to build buy-in. Leaders should highlight instances where the AI’s contributions helped the team succeed, reinforcing its value. Additionally, involve the team in configuring or giving feedback on the AI (e.g., adjusting parameters together) to increase their sense of control and acceptance (Glikson & Woolley, 2020).
- Lmphasise Human Value: Continuously remind and demonstrate to team members that their creativity, expertise, and social skills are irreplaceable. Use the time and insights gained from AI to invest more in those human areas (e.g., schedule brainstorming sessions, personal development opportunities). By doing so, team members see tangible benefits from AI – an increase in meaningful work and personal growth – which counters fears of obsolescence and keeps them engaged (Huang & Rust, 2021).
- Establish Ethical Protocols: Set guidelines for AI usage that ensure ethical considerations are not overlooked. For instance, require human review for any AI-driven decision that significantly impacts people’s careers or clients, and commit to regular audits of AI outcomes for fairness (Zhang, 2021). By formalizing these protocols, organisations reinforce that AIs are tools under human governance, not unchecked decision-makers.
5.4 Limitations and Future Research:
Our framework is conceptual and based on current technology and known case evidence. As such, it might need adaptation for different contexts or more advanced AI capabilities. For example, in highly creative domains (like design or strategy), Domain 2’s operational co-piloting might be less relevant, whereas Domain 1’s cognitive partnership (ideation support) could be key. Conversely, in highly structured operations (like supply chain or call centres), AI might take on larger portions of Domain 2 tasks. Another limitation is that our framework assumes a generally positive organisational climate and tech-savvy teams. Cultural factors and regulatory environments can influence receptiveness to AI leadership. We expect the framework’s effectiveness will vary with such conditions – for instance, in cultures with high uncertainty avoidance, employees might trust AI less (Hofstede, 2001), requiring more intensive trust-building efforts.
Based on our framework, we put forward four initial propositions as a starting point for future empirical study, summarised below:
- P1 (Decision Quality): Teams in which an AI serves as a cognitive partner (providing analysis) and the human leader serves as final decision arbiter will achieve higher decision quality than teams led solely by humans or solely by AI, if the team has calibrated trust in the AI and transparency of AI reasoning as enabling conditions.
- P2 (Efficiency & Agility): Teams in which an AI operational co-pilot manages routine coordination while a human leader handles exceptions will be more efficient (e.g., faster task completion, reduced idle time) and more agile in re-planning after disruptions, compared to teams without AI assistance or teams with AI but lacking human exception management. (This assumes the human leader maintains clarity of authority and intervenes appropriately as needed.)
- P3 (Skill Development & Engagement): Teams that use an AI skills coach alongside human mentoring will show faster individual skill improvement and equal or higher employee engagement than teams relying only on human coaching or only on AI feedback, provided the human leader actively maintains a supportive learning culture. (In other words, AI should supplement rather than replace human mentoring to realise benefits.)
- P4 (Compliance & Trust): Teams where an AI sentinel monitors performance and ethical adherence, with a human leader ensuring accountability and addressing any issues, will exhibit higher compliance with quality and ethical standards – and maintain team trust – compared to teams with purely human monitoring or unguided AI monitoring. (This holds when there is transparency about AI monitoring and consistent human-led ethical enforcement; without those conditions, monitoring by AI could erode trust.)
Future Research:
We encourage empirical evaluation of the framework’s propositions. Researchers could design experiments comparing teams with and without AI co-leaders to measure differences in decision quality or team efficiency (Propositions P1 and P2). Surveys could measure team trust in AI and leader adaptability as mediators of team performance, testing if co-leadership’s success indeed flows through these factors (as we theorise). Qualitative studies could also explore how leaders implement boundaries and build trust (e.g., case studies of organisations that have introduced AI in management). A better understanding of moderators (e.g., task type, team digital literacy, or culture) could refine the framework. The constructs we’ve identified (like role clarity around AI, trust in AI, etc.) provide a foundation for developing measurement scales and hypotheses. Table 2 outlines some key variables and potential measures.
Table 2. Key Constructs for Studying Human–AI Co-Leadership
| Construct (Role or Factor) | Definition (in context of co-leadership) | Example Measure (for future research) |
|---|---|---|
| Co-Leadership Implementation | Extent to which AI is integrated as a co-leader (vs. a mere tool). | Categorical (teams with vs. without AI in leadership role) or continuous index (e.g., % of leadership tasks handled by AI). |
| Team’s Trust in AI (mediator) | Team’s willingness to rely on AI’s outputs appropriately. | Survey scale for trust in AI (adapted from trust in automation scales – e.g., items on reliability, competence, and transparency). |
| Leader Adaptability (moderator) | Human leader’s ability to adjust style/processes when working with AI. | 360° feedback or self-assessment of behaviors (e.g., leader’s openness to AI input, efforts to facilitate use of AI). |
| Team Performance (outcome) | Productivity and decision quality of the team. | Objective KPIs (project completion time, error rates) and supervisory ratings of decision effectiveness. |
| Team Innovation (outcome) | Team’s capacity for creative solutions and improvements. | New initiatives, patents, ideas generated; innovation climate survey ratings. |
| Member Well-Being (outcome) | Employee job satisfaction, engagement, and stress in an AI-augmented team. | Standard engagement and job satisfaction surveys; qualitative feedback on work experience with AI. |
6. Conclusion
This paper introduced a conceptual framework for augmented leadership in human–AI hybrid teams, addressing a significant gap in both theory and practice. By grounding the framework in established research and using real examples, we aimed to ensure it is theoretically sound and practically relevant. The framework breaks down team leadership into four key domains and specifies how AI and human skills can be optimally combined in each. In doing so, it reframes AI from being a potential “boss” or mere tool into a collaborative co-leader – an approach supported by emerging evidence but not previously formalised. Our analysis suggests that, under the right conditions, human–AI co-leadership can outperform traditional models of leadership. Decisions can be more data-driven without losing human judgement, execution can be more efficient without sacrificing flexibility, team members can develop faster while still feeling personally supported, and high standards can be maintained without oppressive monitoring. Crucially, the human leader’s role remains vital – indeed, it becomes more specialised and, in some respects, more demanding, as leaders must be adept at managing AI interactions and championing ethical, human-centric values in tandem with AI use. For organisations, this means AI should not be seen as replacing leaders, but as redefining leadership roles. Companies that manage this integration well – clarifying how AI fits into their leadership processes and training their managers accordingly – are likely to gain in productivity and innovation, as well as attract talent excited to work with advanced technology under inspiring human leadership. Conversely, mismanaging AI integration (either by giving AI too free a rein or by not leveraging AI at all out of fear) could lead to the pitfalls noted in our review, such as diminished trust or a lost competitive edge. We believe that by moving beyond polarising “AI vs. human” debates and focusing on practical co-leadership design, organisations can achieve the “best of both worlds” in team leadership. The contributions of this research are a step toward that goal – offering a pathway to lead in the age of AI not by resisting it, but by embracing it alongside human ingenuity and empathy.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211.
Bass, B. M. (1985). Leadership and performance beyond expectations. New York, NY: Free Press.
Bazerman, M. H. (2020). Better, Not Perfect: A Realist’s Guide to Maximum Sustainable Goodness. New York, NY: HarperCollins.
Bhattacharyya, R., Menon, S., & Ehrlicher, J. (2023). AI in performance management: Mitigating surveillance stress. Journal of Business Ethics, 182(1), 165–183.
Bock, D., Holtgrewe, U., & Kopp, T. (2022). Augmented call center work: Exploring real-time AI coaching. New Technology, Work and Employment, 37(3), 297–312.
Brown, M. E., Treviño, L. K., & Harrison, D. A. (2005). Ethical leadership: A social learning perspective for construct development and testing. Organizational Behavior and Human Decision Processes, 97(2), 117–134.
Brynjolfsson, E., Hitt, L. M., & Kim, H. H. (2011). Strength in numbers: How does data-driven decision-making affect firm performance? Social Science Research Network, Working Paper No. 1819486.
Burns, J. M. (1978). Leadership. New York, NY: Harper & Row.
Chougule, P. (2025, June 29). India leads global AI usage with 92% user adoption in 2025. The Bridge Chronicle.
Crawford, W. S., Laird, M. D., & Moss, J. A. (2023). Will AI change the nature of leadership? MIT Sloan Management Review, 64(2), 14–16.
Davenport, T. H., & Kirby, J. (2016). Only humans need apply: Winners and losers in the age of smart machines. New York, NY: Harper Business.
Eaton, K. (2025, June 30). Salesforce’s CEO just said AI now does half of all company tasks. Inc. (online).
Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383.
Endsley, M. R. (2017). From here to autonomy: Lessons learned from human–automation research. Human Factors, 59(1), 5–27.
Fayol, H. (1949). General and industrial management (C. Storrs, Trans.). London, UK: Sir Isaac Pitman & Sons. (Original work published 1916)
Floridi, L., Cowls, J., Beltrametti, M., et al. (2018). AI4People—An ethical framework for a good AI society: Opportunities, risks, principles, and recommendations. Minds and Machines, 28(4), 689–707.
Fleishman, E. A. (1953). The description of supervisory behavior. Journal of Applied Psychology, 37(1), 1–6.
Glikson, E., & Woolley, A. W. (2020). Human trust in artificial intelligence: Review of empirical research. Academy of Management Annals, 14(2), 627–660.
Goleman, D. (2004). What makes a leader? Harvard Business Review, 82(1), 82–91.
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: LMX theory over 25 years. Leadership Quarterly, 6(2), 219–247.
Hackman, J. R., & Walton, R. E. (1986). Leading groups in organizations. In P. S. Goodman (Ed.), Designing effective work groups (pp. 72–119). San Francisco, CA: Jossey-Bass.
Haenlein, M., & Kaplan, A. (2021). Artificial intelligence and robotics: Shaking up the business world. Journal of Business Research, 124, 405–407.
Herzberg, F. (1968). One more time: How do you motivate employees? Harvard Business Review, 46(1), 53–62.
Hoff, K. A., & Bashir, M. (2015). Trust in automation: Integrating empirical evidence on factors that influence trust. Human Factors, 57(3), 407–434.
Hofstede, G. (2001). Culture’s consequences: Comparing values, behaviors, institutions, and organizations across nations (2nd ed.). Thousand Oaks, CA: Sage.
Huang, M.-H., & Rust, R. T. (2021). A strategic framework for artificial intelligence in marketing. Journal of the Academy of Marketing Science, 49(1), 30–50.
Jaakkola, E. (2020). Designing conceptual articles: Four approaches. AMS Review, 10(1–2), 18–26.
Jarrahi, M. H. (2018). Artificial intelligence and the future of work: Human–AI symbiosis in organizational decision making. Business Horizons, 61(4), 577–586.
Johnson, S. K., Nguyen, B., Groth, M., & White, L. (2020). Can your manager be a machine? The role of artificial intelligence in employee management. Journal of Applied Psychology, 105(9), 1042–1054.
Kapoor, S. (2023, November 1). IBM’s CEO on using AI to reassign hundreds of HR staff. Entrepreneur.
Kamar, E., Hacker, S., & Horvitz, E. (2019). Challenges in deploying AI in team coordination: A multistudy analysis. Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, 7(1), 62–71.
Kotter, J. P. (1996). Leading change. Boston, MA: Harvard Business School Press.
Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80.
Lyons, J. B., Vo, T., & Potochnick, M. (2022). Transparency in autonomous systems: Influence on trust and reliance in human–machine teaming. Applied Ergonomics, 98, 103550.
Madsen, M., & Gregor, S. (2000, December). Measuring human trust in autonomous systems. Proceedings of the 11th Australian Conference on Information Systems (pp. 53–59). Brisbane, Australia.
Martin, K. (2019). Ethical implications and accountability of algorithms. Journal of Business Ethics, 160(4), 835–850.
Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas, E., & Cannon‐Bowers, J. A. (2000). The influence of shared mental models on team process and performance. Journal of Applied Psychology, 85(2), 273–283.
Mesaglio, C. (2025). Top insights from the Gartner AI trust survey. Stamford, CT: Gartner Research.
Mintzberg, H. (1973). The nature of managerial work. New York, NY: Harper & Row.
Nourani, M., Ingram, M., & Mott, B. (2020). The effects of meaningful and meaningless explanations on trust and perceived system accuracy in intelligent systems. Proceedings of the 25th Int. Conference on Intelligent User Interfaces (IUI), 189–201.
OECD. (2019). Recommendation of the Council on Artificial Intelligence. Paris: OECD Legal Instruments (OECD/LEGAL/0449).
Pearce, C. L., & Conger, J. A. (Eds.). (2003). Shared leadership: Reframing the hows and whys of leadership. Thousand Oaks, CA: Sage.
Raisch, S., & Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox. Academy of Management Review, 46(1), 192–210.
Rosenberg, A., Tschopp, D., & Moerschell, L. (2022). Ethical leadership and artificial intelligence in human resource management. Human Resource Management Review, 32(1), 100789.
Rosenblat, A., & Stark, L. (2016). Algorithmic labor and information asymmetries: A case study of Uber’s drivers. International Journal of Communication, 10, 3758–3784.
Seeber, I., Maier, R., De Llano, M. T., Gallenkamp, J., et al. (2020). Machines as teammates: A research agenda on AI in team collaboration. Information & Management, 57(2), 103174.
Sheridan, T. B. (2019). Monitoring behavior and supervisory control of automated systems. In J. D. Lee & A. Kirlik (Eds.), The Oxford Handbook of Cognitive Engineering (pp. 180–190). Oxford, UK: Oxford University Press.
Siggelkow, N. (2007). Persuasion with case studies. Academy of Management Journal, 50(1), 20–24.
Simon, H. A. (1977). The new science of management decision (Rev. ed.). Englewood Cliffs, NJ: Prentice Hall.
Singh, J., Taneja, S., Jacelon, C. S., & Eastman, D. (2021). Employee acceptance of AI-enabled training systems: A psychological empowerment perspective. Journal of Applied Psychology, 106(6), 849–867.
Susskind, D., & Susskind, R. (2022). The future of the professions: How technology will transform the work of human experts (Updated ed.). Oxford, UK: Oxford University Press.
Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367.
Trist, E. L. (1981). The sociotechnical perspective. In A. H. Van de Ven & W. F. Joyce (Eds.), Perspectives on organization design and behavior (pp. 19–75). New York, NY: Wiley-Interscience.
Venkatesh, V., Thong, J. Y., & Xu, X. (2012). Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology (UTAUT2). MIS Quarterly, 36(1), 157–178.
West, D., Milliken, F. J., & Gonzalez, R. (2020). Empowering the human–AI partnership: Keeping people in the loop at work. Journal of Management, 46(7), 1379–1408.
Wilson, H. J., & Daugherty, P. R. (2018). Collaborative intelligence: Humans and AI are joining forces. Harvard Business Review, 96(4), 114–123.
World Economic Forum. (2023). The Future of Jobs Report 2023. Geneva, Switzerland: World Economic Forum.
Yukl, G. (2012). Leadership in organizations (8th ed.). Boston, MA: Pearson
Zhang, J. (2021). Algorithmic decision-making and organisational responsibility. Journal of Business Ethics, 168(2), 111–126.
