AI agents are changing leadership – but how?
In the near future, a global investment fund uses AI agents to analyze investments in a complex market. The system begins to misinterpret risk without anyone noticing, even with executives “in the loop.” Its recommendations appear credible and are followed, resulting in significant losses and turmoil in financial markets.
This imagined scenario raises a fundamental question: What happens when leadership is no longer performed exclusively by humans?
AI agents are moving from tools that support human decisions to systems that can analyze information, coordinate activities, make recommendations and act with increasing autonomy. One thing is becoming clear: AI agents will change both leadership itself and the context in which leadership takes place.
Summary based on a published scientific article by Synnøve Nesse.
From traditional leadership to complexity leadership
Most traditional leadership theories assume that leaders and followers are human. AI agents challenge that assumption. As agents become more autonomous, leadership may emerge from interactions between managers, employees, AI agents and technology providers.
This calls for a complexity leadership perspective, focusing on interactions, interdependencies and adaptation in systems that are constantly changing. This is particularly relevant because neither AI technology nor its organizational consequences can be fully predicted.
Two things are likely to change.
-
AI agents will change leadership itself
When AI agents become part of leadership, they affect how power, influence and responsibility are distributed.
Power may shift between managers and employees, between humans and AI agents, and towards technology owners. As agents recommend actions, prioritize information, coordinate work or act autonomously, they increasingly influence organizational decisions.
Where this will lead is uncertain, making scenario thinking important. AI agents could take over more leadership functions, shifting authority towards technology platforms and their owners. Hybrid leadership could emerge, where humans and AI agents share leadership activities and humans give up some control. Or a more human-centred model could develop, where AI strengthens leaders’ capacity to handle complexity and frees up time for judgement, relationships and sense-making.
These developments may drive new forms of leadership and organizing. Leadership may become more distributed, involving different configurations of managers, employees and AI agents – including at executive and board level.
-
AI agents will change the context for leadership
AI agents will not only become actors in organizations; they will change the context in which managers lead.
Human–AI interaction can be understood as a wicked problem: technological solutions to one problem often create new challenges in practice. AI agents may help organizations manage complexity, information overload and faster decision-making, while simultaneously creating questions about responsibility, competence, trust, dependency and control.
Managers cannot design the perfect human–AI organization in advance. Adapting to consequences and learning quickly from working with AI agents therefore becomes critical.
What does this mean for human leadership?
Two questions become particularly important: what remains key to human leadership, and how should humans lead in these systems?
A question of what: Power, influence and responsibility
Leaders need to understand how AI agents redistribute power, influence and responsibility. Who defines an agent’s objectives? How much authority should it have? And who is responsible when humans and AI jointly shape decisions?
These are leadership questions, not only technology questions. Governance and accountability will become as important as AI’s efficiency gains.
A question of how: Judgement under uncertainty
Leading AI agents require judgement about when and how to rely on them. When should an agent act independently? When should humans intervene or challenge its recommendations?
Keeping executives “in the loop” does not ensure meaningful human control if they accept recommendations, they can no longer adequately challenge. Human judgement under uncertainty therefore does not disappear with AI. As AI reshapes both the execution and context of leadership, it becomes increasingly crucial to leadership across levels.
Three takeaways for managers
- Think of AI agents as organizational actors, not only tools. They can influence decisions and potentially how leadership itself is organized.
- Pay attention to where power and responsibility are moving. Use scenario thinking to prepare for different human–AI relationships.
- Build the capacity to learn and exercise judgement. This is a new field, with no established leadership model. Managers need to experiment, learn and continuously reconsider where AI autonomy ends and human judgement should begin.