A manager begins the day with an AI-generated summary of overnight activity. The system has identified emerging issues, compared them with prior patterns, suggested which ones may deserve attention, drafted possible responses, and prepared information for the teams involved.

Elsewhere in the organization, an AI-enabled system goes further. Within defined permissions, it updates records, initiates workflows, requests additional information, and adjusts routine activity without waiting for a person to issue each instruction.

The work has changed, but the organizational chart may not have.

Managers still hold their titles. Decision rights may look much as they did before. Governance forums still meet. Accountability remains assigned to defined roles. Yet some of the informational, analytical, coordinating, and execution work through which management actually happens is now being performed or supported differently.

That creates a more useful question than whether AI will replace managers:

Who is actually managing when human and AI capabilities participate in the same management process?

This is not a claim that an AI system becomes a manager. It is a question about how management functions once concentrated in formal roles are now being performed across people and systems.

Mary Parker Follett and Henry Mintzberg approached managerial work from very different periods and perspectives. Follett challenged management built primarily around personal command and instead emphasized the situation, coordination, participation, and the relationships through which action becomes possible. Mintzberg studied what managers actually do and showed how far real managerial work differs from the orderly picture suggested by formal descriptions of planning, organizing, coordinating, and controlling.

Neither was writing about today's AI systems. Their work remains useful because AI can redistribute activities inside management without removing the need for judgment, coordination, authority, intervention, and accountable action.

1. Follett: the situation matters more than personal command

Mary Parker Follett wrote about authority in an era when hierarchy and command were dominant features of management practice.

Her argument was more nuanced than simply rejecting hierarchy.

In The Giving of Orders, Follett proposed that managers and employees should study the situation together and discover what she called the law of the situation. Rather than reducing management to one person issuing an instruction to another, she argued for understanding what the circumstances themselves required. She also emphasized that situations continue to evolve, which means an earlier instruction cannot simply be treated as permanently detached from changing conditions.

Follett did not reject authority. Her concern was with arbitrary or overly personalized authority and with the assumption that hierarchical position alone supplied sufficient understanding of what action was required. Her wider distinction between power-over and power-with reinforces that theme, but the more useful point here is the relationship between authority and understanding the situation.

That becomes particularly interesting when AI enters managerial work.

An AI-enabled system may have access to more current evidence, a wider evidence set, or relevant context that the manager has not yet seen. It may identify a change sooner or retrieve information that materially alters the interpretation of what is happening.

Follett's argument suggests that management should not disregard relevant evidence simply because it did not originate with the person holding the highest formal position.

But it does not follow that the AI therefore possesses authority.

The law of the situation is not a theory of machine authority. It is a challenge to management to understand circumstances sufficiently well that authority is exercised in response to what the work requires rather than merely because one person has the power to issue an order.

Under AI conditions, that makes the relationship between evidence, judgment, and authority more important, not less.

2. Mintzberg: management is what managers actually do

Henry Mintzberg challenged a different simplification.

Formal descriptions of management often present managerial work as an orderly sequence of planning, organizing, coordinating, and controlling. Mintzberg's observation of managers found something less tidy.

He described managerial work through interpersonal, informational, and decisional roles. Managers develop relationships, exchange information, negotiate, allocate resources, resolve disturbances, represent their organizations, and make decisions under incomplete and ambiguous conditions. The work is often fragmented, fast moving, interruption-heavy, and highly dependent on communication.

AI is entering exactly this lived structure of managerial work.

It can increasingly support informational work by gathering, summarizing, monitoring, retrieving, and distributing information. It can contribute to decisional work by analyzing alternatives, testing scenarios, identifying conflicts, and preparing recommendations. It can support work surrounding interpersonal roles by preparing communications, maintaining shared context, and identifying dependencies.

That support should not be confused with transferring the social and representative dimensions of managerial roles to a system. Leadership, negotiation, representation, and organizational relationships involve institutional and human dimensions that are not equivalent to drafting a message or assembling evidence.

Where AI-enabled systems also receive operational permissions, they may participate directly in execution.

Mintzberg himself later cautioned against assuming that digital technology fundamentally changes the practice of managing. His argument was that technology can intensify characteristics already present in managerial work, including speed, volume, interruption, and the risk of becoming too remote from the realities of the work.

That provides a useful restraint here. AI may redistribute or accelerate important management activities without eliminating the underlying need to connect information, relationships, judgment, and action.

The important AI-era observation is therefore:

management work can be distributed differently even when formal management roles have not changed.

3. Changing the work changes the management role

Organizations often approach AI initially as a task-allocation question.

Which tasks can the technology perform? Which can be automated? Where can a copilot make an employee more productive?

Those are useful questions, but they become insufficient when AI begins to participate in work through which management itself operates.

Suppose a manager previously spent substantial time gathering operational information, consolidating reports, identifying exceptions, and preparing recommendations for a weekly review. AI may remove much of that effort.

The manager has not necessarily become less necessary. The work that remains may simply become more consequential.

More management effort may now be required to challenge interpretation, reconcile conflicting evidence, understand consequences across functions, decide where intervention is warranted, work through dependencies, and coordinate with people whose responsibilities intersect with the issue.

AI may therefore remove work from a management role while increasing the importance of the management contribution that remains.

Management is not simply the residual category of activities AI cannot yet perform.

The relevant question is what contribution the organization requires once information gathering, analysis, coordination, and execution support are redistributed between people and systems.

That may require a different management role, not merely a smaller version of the old one.

4. Human involvement has to have a purpose

A common response to this uncertainty is to require a human in the loop.

That can be an important control, but human presence alone does not define an effective management arrangement.

A meaningful human role has to specify who is involved, what management contribution that person is expected to provide, when intervention occurs, what context is available, what judgment is required, and how that role connects to the authority needed for action.

Human presence is not the same as meaningful human control.

A person positioned as an approver or control point but lacking authority to approve, stop, override, or escalate effectively is not providing the control the design appears to promise. A manager asked to approve hundreds of low-consequence recommendations may become a ceremonial checkpoint rather than a useful source of judgment. A specialist may understand the operating situation better than an executive and still contribute meaningfully through challenge or escalation rather than by holding the final decision right. An executive may possess authority but receive the issue only after meaningful intervention is no longer possible.

The design question is therefore not simply whether a human participates, but what kind of intervention the management system requires.

In some cases the required contribution is approval. In others it may be challenge, exception handling, escalation, override, conflict resolution, reframing, or the ability to halt an action.

The role has to be positioned early enough, supplied with sufficient context, and connected to the authority or escalation route necessary for the intervention still to have consequence.

Follett helps expose one part of this problem: the situation has to be understood through the evidence and experience relevant to it rather than reduced to hierarchical command.

Mintzberg exposes another: management is accomplished through information, relationships, decisions, communication, and action. Changing those activities changes how managerial work actually happens.

The human role therefore cannot be designed simply as a box in a workflow.

It has to be designed around the management contribution that must occur at that point.

5. Role design must include authority boundaries

Redistributing management work between people and AI-enabled systems also requires clarity about authority.

Technical or analytical capability does not itself confer the right to make an organizational commitment. A system may be capable of identifying a problem, recommending a response, or executing an available action without possessing authority to decide that the action should occur.

The same distinction applies to people. Expertise does not automatically create Decision Authority, and formal authority does not guarantee that the person holding it has sufficient context to exercise it well.

This is where Follett's emphasis on the situation requires an important boundary. Better situational understanding informs judgment. It does not determine who is authorized to commit the organization.

Human-and-AI role design therefore has to make explicit where contribution ends and legitimate organizational commitment begins.

An AI-enabled system may monitor conditions, assemble evidence, recommend a response, or act within clearly delegated limits. A human participant may challenge, approve, intervene, or make the consequential decision. Different arrangements are possible.

What matters is that the boundary is deliberate rather than accidental.

The actor can vary. The governance conditions cannot simply disappear because the actor changes.

6. The AIMR extension: redesigning management around human and AI participation

Follett and Mintzberg help us see why inserting AI into existing work is not simply an automation problem.

Follett directs attention toward the situation, coordination, participation, and the limits of treating personal command as the complete basis for effective action. Mintzberg directs attention toward management as it is actually practiced: information moving through relationships into decisions and action, often under pressure and incomplete understanding.

AI changes who, or what, can perform or support particular activities within that practice.

AIMR extends the question by asking how the management system should be designed when people and AI-enabled systems can participate at different points in sensing, interpretation, recommendation, coordination, decision support, execution, monitoring, and learning.

The objective is to determine what management function must be performed, what capability and governance conditions it requires, and which human, AI-enabled, automated, or hybrid arrangement can satisfy those conditions.

That may mean routine activity handled autonomously within delegated boundaries, AI recommendation followed by human judgment, or human decisions supported by continuous AI monitoring. The arrangement can vary with the work.

What cannot remain ambiguous is the management structure around it. The organization must make explicit who or what may decide, who or what may execute, where intervention can occur, and where organizational accountability remains assigned.

This is also why management redesign should follow changes in the work itself.

If AI removes much of the informational burden from a management role, the answer is not automatically fewer managers, nor is it automatically the same managers doing less work. The role itself may need to change.

The organization should ask what judgment, coordination, intervention, relationship management, and accountability the changed work now requires.

Follett warned against reducing effective authority to personal command. Mintzberg warned against confusing formal descriptions of management with the work managers actually perform. AI gives organizations reason to revisit both.

When AI joins the work, the management question is not simply what humans should keep doing. It is how judgment, coordination, intervention, authority, and action should be arranged across the people and systems now doing the work.

Sources and further reading

• Mary Parker Follett, “The Giving of Orders,” published in Scientific Foundations of Business Administration (1926), later collected in Dynamic Administration: The Collected Papers of Mary Parker Follett.

• Mary Parker Follett, Creative Experience (1924).

• Mary Parker Follett, collected writings on power-with and power-over in Dynamic Administration.

• Henry Mintzberg, “The Manager's Job: Folklore and Fact,” Harvard Business Review (1975; HBR Classic edition 1990).

• Henry Mintzberg, The Nature of Managerial Work.

• Henry Mintzberg, “Managing in the Digital Age: Over the Edge?” (2015).